Development of Hydrogen Electrodes For Alkaline Water Electrolysis

Downloaded from orbit.dtu.
dk on: Jan 30, 2018
Development of Hydrogen Electrodes for Alkaline Water Electrolysis
Kjartansdóttir, Cecilía Kristín; Møller, Per
Publication date:
2014
Link back to DTU Orbit
Citation (APA):
Kjartansdóttir, C. K., & Møller, P. (2014). Development of Hydrogen Electrodes for Alkaline Water Electrolysis.
DTU Mechanical Engineering.
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
i
Preface
This thesis is submitted as partial fulfilment of the requirement for the PhD degree at The
Technical University of Denmark (DTU). The work has been carried out at the Department
of Mechanical Engineering, section of Materials and Surface Engineering, in the time
interval of Nov. 2009 – Dec. 2013, including 10 months of maternity leave. The PhD study
was made under the supervision of Professor Per Møller with financial support from the
Energy Technology Development and Demonstration Program in Denmark (EUDP) (project
number: 63011-0200).
Kgs. Lyngby, 3 December 2013
Cecilía K. Kjartansdóttir
ii
Abstract
The share of renewable energy worldwide is expected to increase by 38% in the period
between 2011 and 2035. Consequently, the share of energy from fluctuating energy sources,
such as wind, sun and hydro, will escalate and new alternatives for energy conversion,
energy storage and load management will be needed. Producing hydrogen via water
electrolysis using surplus, low cost, power from renewables offers the possibility of
increased production capacity and load management with no greenhouse emissions.
Hydrogen is a valuable energy carrier, which is able to contribute to various forms of energy,
such as, production of electricity via fuel cells, fuel for internal combustion engines or gas
turbines, or as a raw material for the production of synthetic fuels via Sabatier or Fischer -
Tropsch process. In some situations it may be suitable to simply inject hydrogen into the
existing natural gas based infrastructure.
Alkaline water electrolysis (AWE) is the current standard (stat of the art) for industrial large-
scale water electrolysis systems. One of the main criteria for industrial AWE is efficient and
durable electrodes. The aim of the present PhD study was to develop electrode materials for
hydrogen production in order to improve the efficiency and durability, and decrease the costs
associated with industrial AWE. The primary effort was reserved to the hydrogen electrodes.
Additionally, a new test setup for efficiency and durability measurements was to be designed
and constructed.
During the present PhD study, new hydrogen electrodes with large electrocatalytic active
surface area were developed. The electrodes were produced by physical vapour deposition
(PVD) of about 20 µm of aluminium onto a nickel substrate followed by thermo-chemical
diffusion and selective aluminium leaching. The obtained electrode surfaces were found to
be highly porous; both at micro- and nano-scale, and surface roughness factors of up to 2300
times that of polished nickel were measured. The electrocatalytic surfaces were characterized
to have unique adhesion to the substrate, which is a critical criterion for industrial
applications. High Resolution Scanning Electron Microscope (HR-SEM) images reveal
highly skeletal structure with pores down to a few nanometres. Half-cell potentiodynamic
polarisation curves, recorded at 25C, 200 mA/cm2, show the electrodes to have 385 mV
lower hydrogen overpotential and 50 mV lower oxygen overpotential, when compared to
polished nickel.
Durability test was carried out in an industrial sized bipolar, non-zero gap AWE stack where
the developed electrodes were applied both as anode and cathode. The stack was operated
with 30 wt.% KOH electrolyte at a maximum temperature of 80C and a pressure of 22 bar.
The duration of the test was about 2 years where the stack was operated for approximately
9000 hours. Comparison of data captured from the first month of operation to data captured
after the durability testing period indicates no significant deactivation/deterioration in
performance of the electrodes during the whole operation period. The stack efficiency was
measured to be 81% (HHV), after the test period, at 200 mA/cm2 and 80 C. It is noted that
iii
the electrolysis test stack was in the development stage and some losses were expected in the
stack itself due to stray currents.
Durability measurements were also carried out in a non-zero gap, single cell electrolysis
setup with 50 wt.% KOH electrolyte, at 120C and 200 mA/cm2 for 1600 hours. The
developed electrodes were applied both as cathode and anode. The cell efficiency was
measured to be 88 % (HHV) during the first two hours of operation. After about 100 hours,
the efficiency had decreased to 84% and was thereafter constant throughout the remaining
test period. The reason for the efficiency drop is proposed to be degradation of the
electrodes, caused by gas-erosion corrosion, together with formation of nickel hydrides in
electrode metal lattice. The material degradation was considerably more severe at the anode
compared to the cathode. The durability single-cell measurements indicate no deactivation of
electrodes after shut-downs.
Microstructure investigations on the PVD Al-Ni diffusion couples at 610C indicate the
diffusion mechanism to be dominated by grain boundary diffusion of nickel-rich phases into
the aluminium-rich PVD structure. The first intermetallic phase formed is determined to be
AlNi3. The phase is observed as small particles in the columnar grain boundaries of the
aluminium structure, after only a few minutes of heat treatment. Due to the high mobility of
aluminium at the annealing temperature, finding nickel and nickel-rich species to be the most
mobile during the heat treatment is highly unexpected and is in contrast with what is stated
in the current scientific literature. Together with the AlNi3 particles observed in the
aluminium residue, only Al3Ni and Al3Ni2 are present in the diffusion layer for up to 30
minutes of heat treatment. Over 30 minutes of heat treatment results in depletion of the
aluminium and formation of highly porous -Al2O3 phase at the top surface. After 2 hours, of
heat treatment only Al3Ni2 and thin layers of AlNi and AlNi3 are observed in the diffusion
zone, with no traces of the AlNi3 phase. Heat treatments performed for longer than 2 hour
result in gradual thickening of the AlNi, AlNi3 and -Al2O3 diffusion layers and grain growth
in the Al2Ni3 phase. The diffusion mechanism can be the key to good properties of the
developed PVD Al/Ni electrodes.
Electrodes produced with shorter time of diffusion, 10-30 minutes, are found to be more
prone to alkaline aluminium leaching and only 4-5 wt.% of aluminium residue is found in
the leached skeletal nickel structure. For the electrodes heat treated for 24 hours, up to 15
wt.% aluminium residue is observed in the skeletal nickel structure. However, leaching of
the PVD Al-Ni structure, after short periods of diffusion, results in formation of cracks
perpendicular to the Ni substrate. The cracks reduce/affect the mechanical strength of the
treated coatings. Electrodes heat treated for short times are found not to be stable under
OER. However, selective aluminium leaching of electrodes heat treated for 24 hours results
in dense, crack free and more mechanically stable/stronger structure. Electrochemical
characterisation on the effect of surface area of the developed electrodes, indicate that the
electrocatalytic activity increases in proportion to the porous layer up to the whole 20 µm
investigated.
With the aim of reducing the production costs associated with the electrode manufacturing,
four process techniques and combinations of these were screened. These were; (1) hot dip
iv
aluminising of nickel followed by thermo-chemical diffusion, (2) direct thermo-chemical

diffusion of aluminium and nickel sheets, (3) aluminium ionic liquid electroplating on a
nickel plate followed by thermo-chemical diffusion and (4) physical vapour deposition of
aluminium onto electroplated sulfamate nickel substrate followed by thermo-chemical
diffusion. Due to the high affinity of aluminium towards hydrogen and oxygen, producing
oxide free AlNi alloy coatings in an inexpensive and simple manner was found to be
challenging. Only the direct diffusion between aluminium and nickel sheets in argon
atmosphere and the PVD Al onto electroplated sulfamate nickel were found to give
promising coatings.
v
Resumé
Andelen af vedvarende energi på verdensplan forventes at stige med 38% i perioden mellem
2011 og 2035. Derfor vil andelen af energi fra fluktuerende energikilder, såsom vind, sol og
vandkraft eskalere, og der vil blive behov for nye alternativer til energi konvertering,
energilagring og load management. Produktion af hydrogen via elektrolyse fra overskydende
elproduktion fra vedvarende energi giver mulighed for øget produktionskapacitet og load
management uden drivhusgasser.
Hydrogen er en værdifuld energibærer, der på mange måder er i stand til at bidrage til
forskellige former for energikilder såsom; produktion af el via brændselsceller, brændstof til
forbrændingsmotorer eller gasturbiner, eller som råstof til fremstilling af syntetiske
brændstoffer via Sabatier eller Fischer–Tropsch processen. I nogle situationer kan det
accepteres blot at injiceres hydrogen i den eksisterende naturgas infrastruktur.
Alkalisk elektrolyse (AWE) er den nuværende standard (state of the art) for store industrielle
vand elektrolyse -systemer. Et af de vigtigste kriterier for industriel AWE er effektive og
holdbare elektroder. Formålet med nærværende Ph.d.-projekt var således at udvikle elektrode
materialer til brintproduktion med henblik på at forbedre effektiviteten og holdbarheden,
samt mindske omkostningerne, der er forbundet med industriel AWE. I forlængelse heraf
blev der udviklet nye test setups for evaluering af effektivitet og holdbarhedstests.
I forbindelse med nærværende ph.d.-projekt blev nye brint elektroder med stor
elektrokatalytisk aktiv udviklet. Elektroderne blev fremstillet ved (PVD) deponering af 20
µm aluminium på et nikkel substrat efterfulgt af termokemisk diffusion og selektiv fjernelse
af aluminium via ætsning. Det nye elektrodemateriale var særdeles porøst på såvel mikro
som nanoskala, og opnåede en aktiv overflade op til 2300 gange større end en poleret nikkel
overflade.
De elektrokatalytiske overflader udviste unik adhæsion til substratet, som er afgørende for en
efterfølgende industriel anvendelse. High Resolution Scanning Electron Microscope (HR-
SEM) billeder bekræftede en robust skelet-struktur med porer ned til nogle få nanometer.
Halv-celle potentiodynamisk recordede polarisationskurver ved 25C og 200 mA/cm2 viste
at elektroderne havde 385 mV lavere brintoverspændings potentiale og 50 mV lavere ilt
overspændings potentiale sammenlignet med poleret nikkel.
Elektrodernes stabilitets test blev udført i en industriel bipolar, ikke zero-gap AWE stak hvor
elektroderne blev anvendt både som anode og katode. Stakken var forsynet med 30 vægt %
KOH elektrolyt og havde en maksimal drifts temperatur på 80C ved et gas tryk på 22 bar.
Varigheden af testen var omkring 2 år, hvor stakken var i drift over 9000 timer.
Sammenligning af driftsdata registreret fra første måneds drift til data registreret efter 9000
timer viser ingen signifikant deaktivering af elektrodernes katalytiske aktivitet. Stakkens
effektivitet blev målt til at være 81% (HHV) efter perioden test på 200 mA/cm2 ved 80C
vi
Det bemærkes, at elektrolyse test stakken var i udviklingsfasen, og nogen effektivitets tab
var at forvente i stakken grundet fejl-strømme.
Elektrodetest målinger blev ligeledes udført i en ikke zero-gap, enkelt celle elektrolyse setup
med 50 vægtprocent % KOH elektrolyt, ved 120C og 200 mA/cm2 i 1600 timer. De
udviklede elektroder blev anvendt både som katode og anode. Cellens effektivitet blev målt
til at være 88% (HHV) i løbet af de første to timers drift. Efter omkring 100 timer var
effektiviteten faldet til 84% og effektiviteten var efterfølgende konstant i den resterende del
af testperioden. Årsagen til reduktionen i effektivitet tillægges nedbrydning af elektroderne,
forårsaget af gaserosion eller korrosion, sammen med en mulig dannelse af nikkel hydrider,
hvad angår katoden. Materialet nedbrydningen var mere omfattende ved anoden i end ved
katoden. Single celle målinger indikerede ingen deaktivering af elektroder under pauser -
uden elektrolyse.
Mikrostruktur undersøgelser af PVD Al-Ni diffusion ved 610C indikerer at diffusion

mekanismen er domineret af en korn grænse diffusion af nikkel-rige faser i den PVD
deponerede aluminium fase. Den første intermetalliske fase der dannes er fastslået til at være
AlNi3. Denne optræder i de søjleformede korngrænser af aluminium strukturen efter kun
nogle minutters varmebehandling. Det er særdeles overraskende og uventet at konstatere at
nikkel og nikkelholdige faser er de mest mobile faser under diffusionsglødningen af den
PVD deponerede aluminium på et nikkel basismateriale. Iagttagelserne er i kontrast til hvad
der fremgår af den videnskabelige litteratur.
Udover at der er AlNi3i korngrænserne af aluminium strukturen, observeres der Al3Ni og

Al3Ni2 i diffusionslaget efter op til 30 minutters varmebehandling. Efter 2 timers
varmebehandling kan dannelsen af en meget porøs  - Al2O3 fase ved den øverste overflade
iagttages. Kun Al3Ni2 og tynde lag af AlNi og AlNi3 observeres i diffusionszone, mendens
AlNi3 fase ikke kan identificeres. Varmebehandlinger længere end 2 timer resulterer i
langsom opbygning af AlNi, AlNi3 og  - Al2O3 fase samt kornvækst i Al2Ni3 fasen. Den
noget særprægede diffusionsmekanisme kan vise sig at være årsagen til de gode egenskaber
af den udviklede PVD Al/Ni elektrode. Elektroder fremstillet med korte diffusions tider, 10-
30 minutter, er konstateret at være lettere at ætse selektivt i alkalisk miljø med et rest
aluminium indhold på 4-5 vægt% i den udætsede skeletal struktur.
For elektroder der varmebehandles i 24 timer, kan der observeres op til 15 vægt % rest
aluminium i det resterende nikkel struktur. Selektiv udætsning af PVD Al-Ni struktur efter
kort tids diffusion, resulterer i dannelse af revner vinkelret på nikkel substratet. Revnerne
formindsker den mekaniske styrke af belægninger. Elektroder varmebehandlet i korte tid
viser sig ikke at være stabile under OER. Selektiv ætsning af aluminium fra elektroder
varmebehandlet i 24 timer viser resulterer derimod i tætte, krakfrie og mere mekanisk stabile
strukturer. Elektrokemisk karakterisering af overfladearealets for de nye udviklede
elektroder viser, at den elektrokatalytiske aktivitet stiger proportional til det porøse lags
tykkere, op til 20 µm.
Med henblik på at reducere produktionsomkostningerne i forbindelse med fremstilling af

elektroder, blev fire mulige procesteknologier, foruden den allerede undersøgte afprøvet.
vii
Disse var: (1) varmaluminering af nikkel, efterfulgt af termokemisk diffusion, (2) direkte
termokemisk diffusion af aluminium folie med nikkel plader, (3) elektrolytisk deponering
af aluminium belægninger fra ioniske væsker på et nikkelsubstrat efterfulgt af termokemisk
diffusion og slutteligt (4) PVD belægninger af aluminium på en elektrolytisk deponeret
sulfamate nikkeloverflade efterfulgt af termokemisk diffusion.
På grund af den høje affinitet af oxygen til aluminium og dermed også smeltet aluminiums
høje affinitet til vand og den efterfølgende høje opløsning af hydrogen i smeltet aluminium
kan fremstilling af AlNi legeringer for elektroder betragtes som en udfordring. Det blev
hurtigt klart at termokemisk diffusion, henholdsvis mellem en aluminium folie og nikkel i en
argon atmosfære og en PVD deponeret aluminium belægning på en elektropletteret
sulfamate nikkel var blandt de mest lovende procesteknologier til fremstilling af skeletal
nikkel.
viii
Acknowledgements
First and foremost, I would like to thank my supervisor, Professor Per Møller, for his
valuable supervision, inspiring ideas, enthusiasm and support throughout the whole project.
His unlimited engagement and passion for science and, moreover, for bridging the gap
between research and industry has been a great motivation for me.
I wish to thank the scientists at Siemens A/S (Scion) for welcoming me to their laboratory
facilities. I particularly want to thank Sune Egelund and Michael Caspersen for their help
with experiments, invaluable discussions and collaboration. I also want to thank Melany
Roeefzaad for introducing me to the cyclic voltammetry technique and for her help and
support during optimisation of the measurement setup. Martin Kalmar Hansen is gratefully
acknowledged for his detailed comments on an earlier draft on this thesis. Kasper Bondo
Hansen is thanked for his contribution to the ionic liquid electroplating experiments.
Jørgen Jensen, Alexander Dierking from the company GreenHydrogen.dk and Lars Yde
from Aarhus University are gratefully acknowledged for making the durability testing of the
electrodes in an electrolysis stack, possible.
I am thankful to all my colleagues in the Section of Materials and Surface Engineering for
their assistance during my PhD, and for making work a pleasant place. Especially I wish to
thank Alexander Elmkvist Barington for his preparation work on the hot dip aluminising
specimens and Malene Kaab for the various cooperation and assistance. Special thanks go to
my great office-mates Svava Daviðsdóttir, Visweswara Chakravarthy Gudla and Rameez Ud
Din, for all the interesting and valuable scientific discussions, support and good company.
In particular I want to thank my dear colleague, Trine Nybo Lomholt, for the thorough
proofreading of the thesis, all the encouraging words and all the help when I needed it the
most. Thank you so much Trine.
Last but not least, my deepest thanks go to my beloved Ole for his endless support and
patience during the whole PhD journey. I especially want to thank him for taking such a
good care of our sweet daughter and our home in my abundant absence during the last
several months.
ix
List of Publications
I. C. K. Kjartansdóttir, L. P. Nielsen, and P. Møller, “Development of durable and
efficient electrodes for large-scale alkaline water electrolysis,” International Journal
of Hydrogen Energy, vol. 38, no. 20, pp. 8221–8231, Jul. 2013.
II. C. K. Kjartansdóttir, M. Caspersen, S. Egelund and P. Møller, “Electrochemical

investigation of surface area effects on PVD Al-Ni as electrocatalyst for alkaline
water electrolysis,” Manuscript to be submitted to Electrochimica Acta.
III. C. K. Kjartansdóttir, A. Hossein, T. Kasama and P. Møller, “Investigations of the

diffusion mechanism of PVD Al and Ni couples at 610°C,” Manuscript to be
submitted.
IV. M. Flyvbjerg, C.K. Kjartansdóttir, M. Caspersen and P. Møller, “Unveiling the secrets
of the Standard Hydrogen Electrode - An inspiration for the on-going development
of hydrogen electrocatalyst,” Manuscript submitted to Journal of the American
Chemical Society, Dec. 2013.
List of Participating Activities

I. C. K. Kjartansdóttir , L.P. Nielsen and P. Møller, “2nd Generation Alkaline
Electrolysis for Hydrogen Production,” Presentation. Danish Metallurgical Society
Symposium, Roskilde, Denmark, 2011.
II. C. K. Kjartansdóttir , L.P. Nielsen and P. Møller, “Electrodes with good durability
for alkaline water electrolysis,“ Poster presentation, World Hydrogen Energy
Conference 2012, Toronto, Canada, 2012.
III. C. K. Kjartansdóttir and P. Møller, “Development of hydrogen electrodes for

industrial scale alkaline water electrolysis,“ Danish Metallurgical Society
Symposium, Lyngby, Damark, 2013.
IV. L. Yde, C. K. Kjartansdóttir, F. Allebrod, M. B. Mogensen, P. Møller, L. R. Hilbert, P.

T. Nielsen, T. Mathiesen, J. Jensen, L. Andersen, and A. Dierking, “2nd Generation
Alkaline Electrolysis,“ Århus University Business and Social Science – Centre for
Energy Technologies, Danmark, 2013.
V. A. B. G. S. W, “Ny produktionsmetode muliggør “stinkende billig” brint,“

Ingeniøren, 2010. [Online]. Available: http://ing.dk/artikel/ny-produktionsmetode-
muliggor-stinkende-billig-brint-114709. [Accessed: 16-Nov-2013].
x
Abbreviations
AWE Alkaline water electrolysis
BSE Back-scatter electron
CAD Computer-aided design (CAD)
CE Counter electrode
CV Cyclic voltammetry
EBSD Electron backscatter diffraction
EDS Energy-dispersive X-ray spectroscopy
EIS Electrochemical Impedance Spectroscopy
EUDP The Energy Technology Development and Demonstration Program in
Denmark
GI Grazing incidence
HER Hydrogen evolution reaction
HHV Higher heating value
HR-SEM High resolution scanning electron microscope
HTAWE High temp alkaline water electrolysis
ICCI Ion channelling contrast imaging
IR Ohmic resistance
KOH Potassium hydroxide
LOM Light optical microscope
MEA Membrane electrode assembly
OCP Open circuit potential
OER Oxygen evolution reaction
PEEK Polyether ether ketone
PEM Polymer electrolyte membrane
PTFE Polytetrafluoroethylene
PVD Physical vapour deposition
RE Reference electrode
SE Secondary electron
SEM Scanning electron microscope
SHE Standard hydrogen electrode
SOEC Solid oxide electrolysis cell
SOEC Solid oxide fuel cell
SPE Solid polymer electrolysis
STP Standard temperature and pressure
TS Thermal spraying
WE Working electrode
XPS X-Ray photoelectron spectroscopy
XRD X-Ray diffraction
xi
Contents
Preface i
Abstract ii
Resumé v
Acknowledgements viii
List of Publications ix
List of Participating Activities ix
Abbreviations x
List of Figures xiv
List of Tables xix
1 Introduction 1
1.1 Background 1
1.2 Objectives of the present study and structure of the thesis 3
2 Fundamentals of Water Electrolysis 6
2.1 The Principle 6
2.2 Thermodynamics 7
2.3 The resistance in the electrolysis cell 9
2.4 Efficiency 11
2.5 Electrocatalysis 12
3 Water Electrolysis Technologies 16
3.1 Alkaline Water Electrolysis 18
3.1.1 Cell components 18
3.1.2 Cell configuration 25
3.2 High temp alkaline water electrolysis (HTAWE) 27

3.3 Polymer Electrolyte Membrane (PEM) electrolysis 28
3.4 Solid Oxide Electrolysis Cell (SOEC) 29
4 Methodology for Testing and Characterisation 30
4.1 Structure and morphology 30
xii
4.2 Composition and phase analysis 31

4.3 Efficiency and durability measurements 31
4.4 Pre‐electrolysis 31
4.5 The three electrode‐ electrochemical cell and IR‐drop 34
4.6 Stability of electrodes 38
4.7 Final remarks 38
5 Development of the Test Setups 39
5.1 First generation test setup 39
5.2 Second generation electrolysis test setup 41
5.3 Second generation half‐cell test setup 42
5.4 Industrial electrolysis stack 44
6 Preliminary Work for the Development of New Hydrogen Electrodes 46
6.1 Material selection 46
6.2 Structural modifications 51
7 Manufacturing of High Surface Area Nickel Coatings 54
7.1 Physical vapour deposition of aluminium onto a nickel plate 54
7.1.1 Introduction 54
7.1.2 Experimental procedure 54
7.1.3 Results and discussions 56
7.2 Hot dip aluminising 58

7.3 Thermo‐chemical diffusion of aluminium and nickel sheets 65

7.4 Aluminium ionic liquid electroplating 67

7.5 Physical vapour deposition of aluminium onto electroplated sulfamate nickel 71

7.6 Conclusions on the manufacturing of a high surface area nickel coating 73

xiii
8 Efficiency and Durability Measurements on the Developed PVD Al/Ni Electrodes
74
8.1 Introduction 74
8.2 Experimental procedure 74
8.3 Results and discussions 75
8.3.1 Half‐cell measurements 75
8.3.2 Durability measurements in an electrolysis cell 78
8.3.3 Electrolysis stack measurements 83
8.4 Conclusions for efficiency and durability testing 86
9 Conclusions 88
Bibliography 91
Appended papers
I. Development of durable and efficient electrodes for large‐scale alkaline water
electrolysis
II. Electrochemical investigation of surface area effects on PVD Al‐Ni as electrocatalyst
for alkaline water electrolysis
III. Investigations of the diffusion mechanism of PVD Al and Ni couples at 610°C
IV. Unveiling the secrets of the Standard Hydrogen Electrode ‐ An inspiration for the
on‐going development of hydrogen electrocatalyst
Appendix
xiv
List of Figures

Fig. 1: Illustration of possible routes for hydrogen produced by water electrolysis [5]. ..................... 2
Fig. 2: Outline of the operating principles of an alkaline water electrolysis cell. ................................ 6
Fig. 3: The thermoneutral voltage (Eth) and reversible voltage (Erev) for the production of hydrogen
in a water electrolysis cell. The system is calculated for water in liquid phase. The image is a
redraw from [10]. ..................................................................................................................... 8
Fig. 4: The electrical circuit analogy of the resistances within a water electrolysis cell. ................... 10
Fig. 5: Elementary reaction steps of HER in alkaline media. .............................................................. 13
Fig. 6: The dependence of the electrocatalytic activity for HER on the metal – hydrogen bond
formed [18] ............................................................................................................................ 14
Fig. 7: Scanning electron micrograph of a Zirfon® Perl 500 UTP diaphragm utilised in this study. The
white particles are ZrO2 powder and the grey mesh is polysulfone. ..................................... 19
Fig. 8: 3D plot of the conductivity of aqueous KOH as a function of temperature and concentration
[90]. Courtesy of Frank Allebrod. ........................................................................................... 23
Fig 9: Illustration of a monopolar stack configuration. S stands for the diaphragm separator [7]. ... 25
Fig 10: Illustration of a bipolar stack configuration. S stands for the diaphragm separator and B for
bipolar electrode [7]. ............................................................................................................. 25
Fig. 11: Illustration of non‐zero‐gap and zer‐gap configuration for AWE systems. Courtesy of
GreenHydrogen.dk. ................................................................................................................ 26
Fig. 12: Outline of the operating principle of a PEM electrolysis cell. ............................................... 28
Fig. 13: Outline of the operating principle of a SOEC. ........................................................................ 29
Fig. 14: Degradation of a high surface area nickel cathode during 93 hours of testing at ‐1256 mV
fixed potential. ....................................................................................................................... 32
Fig. 15: Schematic illustration of a three‐electrode electrochemical cell. ......................................... 34
xv
Fig. 16: A simplified version of the electric equivalent circuit for the an ideal three‐ electrode
electrochemical cell setup, to the left, and a real three‐ electrode electrochemical cell
setup, to the right. ................................................................................................................. 35
Fig. 17: Potentiodynamic measurements performed on a polished nickel with and without current
interrupt IR compensation. .................................................................................................... 36
Fig. 18: Schematic illustration of the parallel connection of a low impedance nickel wire to the
reference electrode in order to reduce noise during electrochemical measurements. ........ 37
Fig. 19: Potentiodynamic measurements on a high surface are nickel electrode. One measurement
is prepared with nickel wire coupled to the reference electrode as illustrated in Fig. 18 and
one is measured with a standard three electrode cell setup. ............................................... 38
Fig. 20: CAD images of the first generation of the electrolysis test setup. ........................................ 40
Fig. 21: The first generation electrolysis cell test setup. A) As designed. B) Modified ...................... 40
Fig. 22: CAD drawings of the second generation electrolysis test cell. .............................................. 41
Fig. 23: The half‐cell measurement setup as assembled and connected to the potentiostat. .......... 43
Fig. 24: The construction of the electrodes inside the PTFE beaker of the half‐cell measurement
setup. ..................................................................................................................................... 43
Fig. 25: CAD image of the construction of the electrolysis stack used for durability testing. Courtesy
of GreenHydrogen.dk. ............................................................................................................ 44
Fig. 26: The bipolar configuration of the electrolysis stack used for durability testing. Courtesy of
GreenHydrogen.dk. ................................................................................................................ 44
Fig. 27: Left: The electrolyser system used for H2‐College. Rigth: The hydrogen storage tank from
H2‐College, ............................................................................................................................. 45
Fig 28: Pourbaix diagrams for cobalt in water at 80 °C. ..................................................................... 47
Fig 29: Pourbaix diagrams for iron in water at 80°C. ......................................................................... 47
Fig. 30: Pourbaix diagrams for nickel in water at 80°C. ...................................................................... 47
Fig. 31: Cobalt specimen after storage in water for a few days shows blue corrosion products on the
surface. ................................................................................................................................... 48
Fig. 32: Accumulated average weight loss of nickel and iron in 150°C 30 wt.% KOH at 5 bar O2
pressure. ................................................................................................................................ 49
xvi
Fig. 33: Accumulated average weight loss of nickel and iron in 150°C 30 wt.% KOH at 5 bar H2
pressure. ................................................................................................................................ 49
Fig. 34: Images of iron (to the left) and nickel (to the right) untreated and after 35 weeks of
exposure in 150°C 30 wt% KOH at 5 bar O2. .......................................................................... 50
Fig. 35: Images of iron (to the left) and nickel (to the right) untreated and after 22 weeks of
exposure in 150°C 30 wt% KOH at 5 bar H2 ........................................................................... 50
Fig. 36: The as‐plated platinum black surface. ................................................................................... 51
Fig. 37: High resolution scanning electron microscope images of platinum black surface. .............. 52
Fig. 38: Left: Cross section back scatter electron micrograph of a Raney nickel coating prepared by
thermal spraying of Al/Ni powder onto a nickel substrate. Right: Schematic illustration of a
thermally sprayed structure [42]. .......................................................................................... 53
Fig. 39: PVD aluminium deposited nickel substrate, as received, heat treated and alkaline leached.
................................................................................................................................................ 55
Fig. 40: BSE micrographs of nickel substrate, left: surface, right: cross section. ............................... 56
Fig. 41: SE micrograph of the PVD aluminium in as deposited state in different magnifications. ..... 56
Fig. 42: Left: Ni‐Al binary alloy phase diagram from [137]. The horizontal line indicates the thermo‐
chemical diffusion temperature (610C). Right: Cross section of a PVD Al/Ni electrode after
heat treatment at 610°C for 24h, prior to leaching. The arrows indicate the supposed
intermetallic phase found in the cross section. The numbers refer to the EDS analysis in
Table 7. ................................................................................................................................... 57
Fig. 43: Cross section SEM micrographs of a PVD Al/Ni electrode after heat treatment at 610°C for
24h, followed by alkaline aluminium leaching. Left: Prepared by the first leaching procedure
(PVD Al/Ni 1). Right: Prepared by the second leaching procedure (PVD Al/Ni 2). ................ 58
Fig. 44: The clay crucible used for the hot dip aluminising as placed in the furnace before and after
melting of the aluminium....................................................................................................... 59
Fig. 45: Two hot dip aluminized nickel coupons. A) Without purging of argon gas and B) with
purging of argon gas. ............................................................................................................. 60
Fig. 46: Cross section LOM image of a nickel plate immersed in molten aluminium. Left: for 2 sec.
Right: for 8 sec. ...................................................................................................................... 61
xvii
Fig. 47: Cross section BSE SEM micrographs of a nickel coupon after hot dip aluminizing for 30
seconds. ................................................................................................................................. 62
Fig. 48: Al‐Ni binary alloy phase diagram [137]. ................................................................................. 63
Fig. 49: Cross section LOM micrograph of a nickel coupon after 5 minutes of hot dip aluminising. . 64
Fig. 50: Cross section SEM micrographs of a nickel coupon hot dip aluminised for 30 sec. and heat
treated at 610°C for 24 hours. ............................................................................................... 64
Fig. 51: Cross section BSE SEM micrographs of the thermo‐chemical diffused aluminium and nickel
sheets. .................................................................................................................................... 66
Fig. 52: Cross section BSE SEM micrograph of the thermo‐chemical diffused aluminium and nickel
sheets after selective leaching of aluminium. ....................................................................... 67
Fig. 53: Schematic illustration of the experimental setup for the aluminium ionic liquid
electroplating procedure. ...................................................................................................... 68
Fig. 54: Cross section LOM micrographs of ionic liquid electroplated Al on a nickel substrate. Left, as
plated. Right, heat treated for 24 h. at 610°C. ....................................................................... 69
Fig. 55: Cross section SEM micrograph of a ionic liquid electroplated aluminium on a nickel
substrate after 24 hours heat treatment at 610 °C. .............................................................. 69
Fig. 56: Cross section SEM micrographs of PVD aluminium on sulfamate nickel substrate heat
treated at 610°C for 4 hours. ................................................................................................. 72
Fig. 57: Cross section SEM micrographs of PVD aluminium on sulfamate Ni substrate heat treated at
610°C for 4 hours and alkaline aluminium leached. .............................................................. 72
Fig. 58: Cathodic potentiodynamic polarisation curves recorded on the developed electrocatalyst
produced with different heat treatments (10, 20, 30 minutes and 24 hours) compared to
polished nickel. The electrolyte contains ............................................................................... 75
Fig. 59: Re‐plot of the cathodic potentiodynamic curves from fig. Fig. 58 in order to find I0 and Tafel
slopes. The grey lines indicate the Tafel slope 1 from 0.02 to 0.1 A/cm‐2. ............................ 76
Fig. 60: Anodic potentiodynamic polarisation curves recorded on the developed electrocatalyst
compared to polished nickel. ................................................................................................. 77
Fig. 61: Re‐plot of the anodic potentiodynamic curves from Fig. 60 in order to find I0 and Tafel
slopes. The grey lines indicate the Tafel slope 1 from 0.02 to 0.1 A/cm‐2. ............................ 78
xviii
Fig. 62: Electrolysis whole cell measurements, where the developed electrode is applied both as
anode and cathode, operated at 200 mA/cm2, 120C in 50%KOH. Electrolysis
measurements operated under the same conditions, where polished nickel is applied as
anode and cathode, is plotted for comparison. ..................................................................... 79
Fig. 63: Left: Black particles found in the electrolyte after over 1600 time of durability testing. Right:
the electrolyte in the electrolysis cell after short duration of testing with a new electrolyte.
................................................................................................................................................ 80
Fig. 64: The PVD Al/Ni electrodes used for single cell electrolysis durability testing after over 1600
hours of operation. Left: Cathode. Right: Anode. The electrode surfaces still appear black
indicating that some remaining high surface area skeletal nickel coating. ........................... 81
Fig. 65: Cathodic potentiodynamic polarisation curves recorded on the PVD Al/Ni electrode applied
as cathode for over 1600 hours durability testing. The electrolyte contains 1 M KOH and the
experiments are performed at 25C. ..................................................................................... 82
Fig. 66: Re‐plot of the cathodic potentiodynamic curves in Fig. 65 in order to find I0 and the Tafel
slope. The dotted lines indicate the Tafel slope 1 from 0.01 to 0.1 A/cm‐2. .......................... 82
Fig. 67: Current density vs. cell voltage recorded on the 17 cell electrolysis stack during the first
month of operation. The operation conditions were 40‐45C and 10 bar. ........................... 84
Fig. 68: Current vs. voltage and efficiency data captured on the 17‐cell bipolar electrolysis stack
after approximately 9000 operating hours. The stack was operated at 60C and 22 bar. The
efficiency calculations are based on the HHV. ....................................................................... 85
Fig. 69: Temperature vs. voltage plot captured on the 17‐cell bipolar electrolysis stack at 22 bar and
200 mA/cm2current density, after approximately 9000 operating hours, showing the
temperature dependence of the cell voltage. The efficiency calculations are based on the
HHV. ....................................................................................................................................... 85
xix
List of Tables

Table 1: An overview of the main current electrolyser manufacturers, their product and
performance data. The table is adopted from [46]. .............................................................. 17
Table 2: Half‐cell reactions and charge carriers for the three main types of water electrolysis [49] 18
Table 3: An overview of the hydrogen overpotential for some electrocatalysts for AWE found in the
literature. ............................................................................................................................... 21
Table 4: An overview of the oxygen overpotential for some electrocatalysts for AWE found in the
literature. ............................................................................................................................... 22
Table 5: Contaminant residues in analytical clean KOH from Merck that could influence the
electrocatalitic activity of the electrode during testing. ...................................................... 33
Table 6: External residues found with XPS measurements performed on high surface nickel
electrodes with three different pre‐treatments. ................................................................... 33
Table 7: Results from the cross section EDS analysis on PVD Al/Ni electrodes before and after the
first and the second leaching procedure. The phase numbers refer to the numbers in Fig. 42
and Fig. 43. All elements from the periodic table are analysed. ........................................... 57
Table 8: Experimental series for hot dip aluminizing ......................................................................... 59
Table 9: Results from EDS analyse on the hot dip aluminised specimen, prior to heat treatment (Fig.
47) and after 24 h. of heat treatment (Fig. 50). All elements from the periodic table except
for carbon are analysed. ........................................................................................................ 63
Table 10: EDS analyses on thermo‐chemical diffused aluminium and nickel sheets (Fig. 51 and 52).
All elements from the periodic table except for carbon are analysed. ................................. 66
Table 11: EDS analyses on the aluminium ionic liquid electroplated nickel substrate (Fig. 54 and 55).
All elements from the periodic table except for carbon are analysed. ................................. 69
Table 12: EDS analyses on the PVD Al on sulfamate Ni specimens (Fig. 56and Fig. 57). All elements
from the periodic table except for carbon are analysed. ...................................................... 72
Table 13: Summary from screening of process techniques for producing high surface nickel
electrocatalyst ........................................................................................................................ 73
Table 14: Tafel slopes, HER overpotential (ηHER) and calculated efficiency (ηref) from the cathodic
potentiodynamic measurements recorded on the developed electrode. ............................ 76
xx
Table 15: Tafel slopes, OER overpotential (ηOER) and calculated efficiency (ηref) from the anodic
potentiodynamic measurements recorded on the developed electrode. ............................ 78
Table 16: Efficiency calculations for electrolysis durability test, operated at 200 mA/cm2, 120C and
50 wt% KOH, recorded on the developed electrodes compared to polished nickel. The
efficiency values are calculated according to the HHV. *Measured immediately after change
of electrolyte. ......................................................................................................................... 79
Table 17: Results and calculated efficiency (η) from the cathodic potentiodynamic measurements
recorded on electrode applied as cathode for over 1600 hours single cell electrolysis
testing. *Calculated according to the theoretical potential of HER ( ‐943 mV vs. Hg/HgO). . 83
Table 18: Selected data from Fig. 67, Fig. 68 and Fig. 69 for durability assessment. The efficiency
calculations are based on the HHV. ....................................................................................... 86
Introduction 1
1 Introduction
1.1 Background
According to the World Meteorological Organization (WMO), the concentration of

greenhouse gases in the atmosphere rose in 2012 to the highest ever recorded [1]. Carbon
dioxide, mainly from fossil fuel related emissions, is responsible for about 80% of the
warming effect from greenhouse gases. The increase in carbon dioxide in the atmosphere
from 2011 to 2012 was measured to be higher than the average growth rate for the past ten
years. This happens despite the plentiful goals and policies towards reduction in CO2
emission worldwide counting, for instance;
 The target of limiting the long – term global temperature increase to 2C as agreed at
the United Nations Framework Convention on Climate Change Conference, Mexico
2010 [2].
 The energy and climate change objectives for 2020 accepted at the European
Council in 2007 with the aim of 20% reduction in greenhouse emission and 20%
increase in the share of renewable energy together with commitment of 80-95%
reduction in greenhouse emission by 2050 [3].
Additionally, according to the World Energy Outlook released in November 2013 [4], the
worldwide energy demand will increase by one-third from 2011 to 2035. Accordingly, in
order to meet the ever increasing energy demand and diminish the risk of dramatic climate
change in the coming years, rapid shift in the global energy trend away from coal is
necessary.
Renewable energy is commonly recognised to be the answer for a more secure, reliable and
sustainable future. The share of renewables in primary energy worldwide is expected to
increase by 38% from 2011 to 2035 [4]. This implies that a larger share of the energy will
come from fluctuating energy sources such as wind, sun and hydro. Hence, developing new
efficient alternatives for energy conversion, energy storage and load management is
essential.
Combining production of hydrogen via water electrolysis with renewable energy sources
offers the possibility of increased production capacity and load management with no
greenhouse emissions. Hydrogen as an energy carrier comprises the advantage of flexibility,
being able to contribute to various kinds of energy sources. Fig. 1 shows possible routes for
hydrogen produced by water electrolysis. The most obvious route is the production of
hydrogen with excessive power from fluctuating renewables followed by storage and later
production of electricity via fuel cells, combustion engines or gas turbines, when needed.
The hydrogen can also be used as a raw material for production of synthetic fuels, such as
2
m
methane by thhe Sabatier process
p [153]], liquid fuels by the Fisscher–Tropscch synthesis [154]
orr simply injeccted into the existing natuural gas infraastructure [155].
Figg. 1: Illustration
n of possible rouutes for hydrogen produced by
y water electrolyysis [5].
A variety of water
w electrolysis system
ms have been proposed deeveloped andd constructed d over
thhe years. Am mong the co ommercially available water
w electro
olysis systemms, alkaline water
eleectrolysis (A
AWE) comprrises by far tthe highest production
p rate and the llowest produ uction
coost, making the techniqu ue the currennt standard for large-scaale water eleectrolysis syystems
[10,15]. Howeever, finding g low cost eleectrode mateerials that are both efficiient and havee long
terrm stability is
i one of the remaining cchallenges wiithin the field
d of AWE.
Introduction 3
1.2 Objectives of the present study and structure of the thesis
The present PhD study is a part of a research, industrial, development and demonstration
program called 2nd Generation Alkaline Electrolysis. The project was initiated in 2008 and
the project participants were Århus University Business and Social Science – Centre for
Energy Technologies (CET (former HIRC)), Technical University of Denmark – Mechanical
Engineering (DTU-MEK), Technical University of Denmark – Energy Conversion (DTU-
EC), FORCE Technology and GreenHydrogen.dk. The project was supported by EUDP.
Expected results from the project were as follows:
A generation of alkaline electrolysers characterized by:

 Increased electrode efficiency to more than 88% (HHV) at a current density of
200 mA /cm2.
 Increased operation temperature to more than 100C
 Operation pressure of more than 30 bar
 Improved stack architecture that will decrease the price of the stack with at least
50%.
 A modular design that makes it easy to customize plants in the size from 20 to
200 kW.
 Demonstration of a 20 kW 2nd generation stack at H2-College at the campus of
Århus University in Herning.
The overall purpose is research, development and demonstration of the emerging renewable
energy concept based on hydrogen as a renewable energy carrier, produced from water and
wind power. The methodology of the project was to be implemented through 8 different
work packages one of these being improvement of electrodes.
There are two ways to increase the efficiency of an electrode. Increase the electrocatalytic
properties of the electrode surface or increasing the specific surface area of the electrode.
Both of these results in a reduction of the necessary voltage to draw the current applied. The
work package will deal with both, through literary studies and laboratory tests in order to
find and specify for production the optimal electrode solution regarding efficiency, durability
and price.
The objective of the present PhD study was to identify electrode materials for hydrogen
production in order to improve the efficiency and durability and decrease the costs associated
with alkaline water electrolysis. New materials were to be designed for the electrodes
containing no precious metals. The primary focus of the study was reserved to the hydrogen
electrodes (the cathodes). Additionally, new test setup for efficiency and durability
measurements was to be developed and constructed.
4
The thesis is divided in to 9 chapters and 4 appended papers.
Chapter 1 is a short introduction of the drive for producing hydrogen via water electrolysis
together with the objectives of the PhD study and structure of the thesis.
Chapter 2 encloses the fundamentals of water electrolysis including the main principle, the
thermodynamics and an overview of the main cause for an efficiency loss in an electrolysis
cell. The main principle behind the kinetics of a hydrogen electrode is introduced in a
subchapter called Electrocatalysis.
Chapter 3 gives an overview of the current main types of water electrolysis technologies.
The chapter will be focused on the current status and research trends of alkaline electrolysis.
In Chapter 4 the methodology for testing and characterisation of the developed electrodes
surfaces is covered. The thesis is written with the intentions of assisting the next generation
of researchers in the field of alkaline water electrolysis electrocatalysts. Well reported
principles behind the measurements and analytic methods used will therefore not be
described. Instead, the emphasis will be on experimental challenges and principles,
encountered during the PhD study, that are not found to be well covered in the literature.
In Chapter 5 the development process of the electrolysis test setups constructed during the
present project will be presented.
In Chapter 6 the practice for material selection and structure modification for developing a
new hydrogen electrocatalyst for alkaline electrolysis is introduced.
In Chapter 7 a screening of process techniques and combination of these for producing high
surface area nickel electrodes for alkaline electrolysis will be described and the main results
reported.
In Chapter 8 results from electrochemical and durability tests prepared on the most
promising electrocatalytic surface from the screening in chapter 7 are reported.
Overall conclusions of the findings reported in the thesis and the appended papers, are given
in Chapter 9.
Appended papers:
Three of the appended papers are first-authored by the candidate and for the fourth paper the
candidate is a co-author.
P-1: Development of durable and efficient electrodes for large-scale alkaline water
electrolysis.
In this paper, studies on the electrode developed in this PhD study are reported. Structural
characterisation of the electrodes is performed by high-resolution scanning electron
microscope. The electrocatalytic activity of the developed electrodes is studied with steady-
Introduction 5
state electrochemical measurements and cyclic voltammetry. Durability tests are carried out
in an industrial scale-electrolysis stack.
P-2: Electrochemical investigation of surface area effects on PVD Al-Ni as

electrocatalyst for alkaline water electrolysis.
In this paper, the influence of the actual surface area on the electrocatalytic activity of the
developed electrocatalyst is investigated. The structure and composition of the developed
electrodes is characterised by the means of a scanning electron microscope and X-ray
diffraction. The electrocatalytic activity and actual surface area are studied with
potentiodynamic polarisation, cyclic voltammetry and electrochemical impedance
spectroscopy.
P-3: Investigations of the diffusion mechanism of PVD Al and Ni couples at 610°C.

In this paper, investigation on the diffusion mechanisms of the developed electrocatalyst is
reported. PVD Al-Ni diffusion couples, heat treated at 610C for few minutes up to 24 hours,
are investigated by the means of, high resolution scanning electron microscope, energy
dispersive X-ray spectroscopy, X-ray diffractometry, electron backscatter diffraction, ion
channelling contrast imaging and transmission electron microscopy.
P-4: Unveiling the secrets of the Standard Hydrogen Electrode - An inspiration for the
on-going development of hydrogen electrocatalyst.
In this paper, a new perspective on electrode design and electrodeposition of the platinum
black electrode are presented.
6
2 Fundamentals of Water Electrolysis

2.1 The Principle
Water electrolysis is a process where electricity is used to decompose water into its
components - gaseous hydrogen and oxygen according to.
1 (1)
H O→H O
2
A typical water electrolysis cell consists of two electrodes, a diaphragm, an electrolyte and a
power supply. The most common type of water electrolysis cell, the alkaline electrolysis cell,
is illustrated in Fig. 2.
Fig. 2: Outline of the operating principles of an alkaline water electrolysis cell.
When current is applied to the system, hydrogen gas is formed at the cathode and oxygen gas
at the anode. The diaphragm is a membrane that hinders the mixing of gasses developed at
the two electrodes and allows the ions to pass. The gas produced can be captured in their
pure form when drifting up from the electrolyte. The reactions taking place at the electrodes,
i.e. the half-cell reactions, differ between the types of water electrolysis techniques applied.
These mechanisms are introduced in chapter 3.
Fundementals of Water Electrolysis 7
2.2 Thermodynamics
As indicated by the power supply in Fig. 2, the decomposition of water into hydrogen and
oxygen is not thermodynamically favourable, i.e. the decomposition process is energy
consuming. The total energy needed for decomposing one mole of water into oxygen and
hydrogen corresponds to the enthalpy change (ΔH) for the reaction, which is 285.9 kJ at
standard conditions [6]. At low temperatures the majority of this energy is required to be in
the form of electrical energy and the rest can be applied as thermal energy. The energy
needed for the decomposition of water can be expressed from the enthalpy changes for the
reaction as follows:
ΔH ΔG TΔS (2)
Where ΔG is the Gibbs free energy change for the reaction, ΔS is the entropy change and T
is the temperature of the reaction in Kelvin. The Gibbs free energy can be regarded as the
minimum amount of the electric energy that has to be applied to the system for the reaction
to take place. ΔG can be calculated from the reversible voltage (Erev) (also called the
equilibrium voltage (Eeq)) according to:
∆G nFE (3)
Where n is the number of moles of electrons transferred in the reaction (here 2) and F is the
Faradays constant (9.64853399 x 104 C/mol). Erev is the absolute minimum voltages needed
in order to produce hydrogen and oxygen via water splitting. Erev, thus equals the sum of the
reversible voltage (Erev) of the anodic and cathodic reaction in the system to:
E E E (4)
At standard conditions the reversible voltage of oxygen and hydrogen are 1.229 V and 0.000
V, respectively, resulting in Erev = 1.229 V [7]. Accordingly, for the following reaction in eq.
(1) to take place in a water electrolysis cell, 1.229 volts must be applied between the anode
and cathode. When inserting the reversible voltage into eq. (3) the Gibbs free energy of the
reaction can be calculated giving ΔG = 237.2 kJ/mol. The electrical potential needed for
decomposition of water for conditions varying from the standard conditions can be derived
directly from the Nernst equation according to [8]:
(5)
E , E ln )
Where R is the gas constant (8.3144621 J/ mol K), PH2 the partial pressure of hydrogen, PO2
the partial pressure of oxygen and PH2O the partial pressure of water. Assuming the same
partial pressure over the whole system, eq. (5) can be written as:
RT (6)
E , , E ln P
nF
8
Consequently, if the system is pressurised, Erev,T,P > Erev, then more electrical energy is
required for the reaction in eq. (1) to take place. The increase in voltage is, however, minor
at low temperatures. If, as an example, an electrolysis cell is pressurised to 200 atm. at 80C,
the increase in Erev is only 0.04 V. Furthermore, it has been shown that pressurising the
system can help in reducing the ionic resistivity in the cell, see concentration overpotential in
section 2.3. When knowing the Gibbs free energy and the enthalpy for the system the energy
that can be applied to the reaction in the form of heat can be calculated from eq. (2). At
standard conditions the thermal energy is:
285.9 kJ⁄mol 237.2 kJ⁄mol 48.7 kJ⁄mol (7)
However, if this amount of heat cannot be integrated into the process, the shortage of energy
has to come from the electrical source, i.e. more than Erev = 1.229 volt difference is required
between the anode and cathode for the water splitting reaction to take place. The total
electrical energy required for maintaining an electrochemical reaction without generation or
absorption of heat, at a specific temperature, is called the total thermo neutral voltage (Eth).
The minimum amount of energy needed for the reaction is equal to the enthalpy of the
thermo-neutral voltage and can be defined as:
ΔH G TS (8)
E ≡
nF nF nF
At standard temperature and pressure (STP) the Eth is calculated to be 1.481 V [9]. Fig. 3
illustrates the thermo-neutral voltage and the reversible voltage for hydrogen production via
water electrolysis as a function of temperature.
Fig. 3: The thermoneutral voltage (Eth) and reversible voltage (Erev) for the production of hydrogen in a water
electrolysis cell. The system is calculated for water in liquid phase. The image is a redraw from [10].
If an electrolysis cell is operated above the Eth conditions, the system generates heat
(exothermic). If, on the other hand, operated below Eth the system absorbs heat
(endothermic). Operating below the Erev voltage the decomposition of water is
thermodynamically impossible.
In Fig. 3 the benefits of producing at elevated temperatures is evident. By increasing the

temperature, a larger amount of the energy needed for the process to occur can be applied in
the form of thermal energy. Therefore, if low cost thermal energy is available the cost
associated with the production of hydrogen via water electrolysis can be minimised. In fact,
at temperatures above 2000°C water can be decomposed directly via thermochemical
processes [11]. Nevertheless, with direct heating alone, without the use of catalysts, much
higher temperatures are required. According to the thermodynamics, temperatures above
4100C are needed for the splitting process to be thermodynamically favourable. In [7] Rand
et al. writes that only ~1vol% of water is decomposed at 2000°C. Whereas in [12] Isao Abe
reports that 5000 K direct heating is needed for full decomposition of water and 2500 K for
partial decomposition.
2.3 The resistance in the electrolysis cell
Apart from the theoretical energy consumption during electrolysis, there are a number of
additional electrical barriers that needs to be overcome for the electrolysis process to occur.
Thus, the cell voltage during electrolysis is always higher than Eth derived from the
thermodynamics. The additional voltage needed to overcome these barriers is often called
overpotential or overvoltage. The overpotentials in an electrolysis cell can be divided into
three categories [13].
 Resistance overpotential (ηres)

 Concentration overpotential (ηcon)
 Activation overpotential (ηact)
The resistance overpotential represents the electrical resistance in the cell from external
wiring, electrical connections to the electrodes and the resistance in the electrodes
themselves. This type of overpotential can usually be minimised by selecting good
electrically conducting materials for the wires and connections and by assuring sufficient
cross sections of those. The resistance in the electrodes themselves can become considerably
large. This can for example be the case for electrodes containing passivated oxides or
isolating impurities, thus, this should be avoided.
The concentration overpotential is caused by the resistance in the ionic transfer in the
electrolyte located between the anode and cathode. The concentration overpotential depends
on the ionic conductivity of the electrolyte, the distance between the anode and cathode, the
conductivity of the diaphragm and the presence of gas bubbles in the electrolyte. The
conductivity of the electrolyte and the cell configuration for AWE cells are discussed in
more detail in section 3.1.
10
During water electrolysis hydrogen and oxygen gas is formed on the electrode surfaces. The
gas bubbles formed are small and do not have the required volume to drift away from the
electrode surfaces immediately. Only after sufficient coalescence the gas bubbles get large
enough to drift away from the electrode surfaces into the electrolyte. The bubble coverage on
the electrode surfaces and the bubble dispersion in the electrolyte are often referred to as the
bubble phenomenon in AWE. The gas bubbles attached to the surfaces block some part of
the active electrode area and thereby hinder the electrochemical reaction to take place. The
gas bubbles in the electrolyte itself evidently increase the ionic resistivity of the electrolyte.
These two effects cause a high ohmic drop during operation and are responsible for the
largest amount of the concentration overpotential in AWE cells [14]. Pressurising the
electrolysis stacks, assuring good convection in the electrolyte and the use of zero gap cell
configuration (see section 3.1.2.2) can be applied for minimising this type of overpotential.
The activation overpotential represents the activation energies of the electrochemical

reactions taking place on the anode and cathode which increase logarithmically with the
current density [15]. The concentration overpotential together with the activation
overpotential is responsible for the greatest deal of the overpotentials during electrolysis. The
activation overpotentials for the oxygen and hydrogen evolution reaction are discussed in
detail in section 2.5.
The overall cell voltage for a water electrolysis cell (under adiabatic conditions) can then be
written as:
E E E E E (9)
Both the concentration overpotential and the resistance overpotential cause heat generation
in the system. Some of the generated heat can, in well isolated electrolysis system, be used
for heating up the electrolyte. In that way the efficiency loss from the overpotentials can be
minimised.
For the ease of identifying every single overpotential source in an electrolysis cell, the
overall cell resistance can be expressed by the electrical circuit analogy as displayed in Fig.
4. Here RE represents the electrical resistance from wiring and connections to the electrodes.
Ranode and Rcathode are the overpotentials required to overcome the activation energies of the
oxygen and hydrogen formation, respectively. RO2 and RH2 is the resistance from the oxygen
and hydrogen bubbles in the electrolyte and on the electrodes surfaces. Rions represents the
resistance in the electrolyte and Rdiaphragm is the diaphragm resistance.
Fig. 4: The electrical circuit analogy of the resistances within a water electrolysis cell.
2.4 Efficiency
The efficiency of the water electrolysis process can be expressed in many different ways,
depending on how the electrolysis system is assessed and compared. As a result, this
sometimes introduces some confusions and misunderstandings in the literature. The cell
efficiency is the most essential for electrode development. The cell efficiency (η) is
calculated by comparing the measured cell voltage either to the reversible or the thermo
neutral voltage as shown in eq. (10) and (11):
E (10)
η
E
E (11)
η
E
If, for instance, an electrolysis cell is operated at STP at cell potentials of Ecell =1.48 V, the
efficiency is calculated to be 100% according to eq. (11). If on the other hand eq. (10) is
used the efficiency is calculated to be only 83%. Thus, when reporting electrolysis efficiency
it is important to inform about how the efficiency is calculated. If not, the reported value is
of no use for the reader. Eth is also known as the higher heating value (HHV). Thus, in the
literature ηth is commonly referred to the efficiency according to the HHV. The ηrev is always
less than 1, because hydrogen cannot be produced if the potential is less than Erev applied to
the system. In contrast, ηth can be higher than 1. This is because some of the heat needed for
the reaction can be absorbed from the environment.
Another way to calculate the efficiency is by comparing the energy input to electrolysis
system with the hydrogen production rate [10]. This can be done according to:
V m h (12)
η
Uit kJ
Where V is the hydrogen production rate at a unit volume of an electrolysis cell, U is the cell
voltage, i is the current and t is the time.
For commercial electrolysis systems the efficiency is typically calculated from the total
energy consumption for each cubic meter of hydrogen produced. According to the HHV,
100% efficiency is reached with 3.54 kWh/Nm3. Thus, ideally, 39 kWh of electricity and 8.9
liters of water are required to produce about 11000 litres or 1 kg of hydrogen at STP [16].
However, the current most efficient commercial electrolysis systems only have a maximum
efficiency of 82% corresponding to 4.3 kWh/Nm3. Performance data for electrolysers from
the leading current electrolyser manufactures is gathered in Table 1 in chapter 3.
12
2.5 Electrocatalysis
An electrocatalyst is a material which can provide low activation pathways for a specific
electrochemical reaction and permit the reactions to occur at high current densities. The
catalytic activity of an electrocatalyst depends on the electron configuration of the catalyst
material (the intrinsic properties) and the structure and geometry (the actual surface area) of
the catalyst. The activity can be measured from the activation overpotential of the catalyst
towards the reaction. The activation overpotential is caused by the resistance against the
reaction itself at the electrode-electrolyte interface. That is, for any chemical or
electrochemical reaction there is an energy barrier, a minimum energy above the average,
which the reactants must possess for the reaction to proceed [13]. The rate determining step
can be either ion or electron transfer across the interface, or it can also be some kind of
conversion of a species involved in the reaction. At a sufficient overpotential and in the
absence of mass transfer limitations the relationship between the current density (i) and the
activation overpotential (ηact) is given by the Tafel equation eq. (13). [17]:
αnF (13)
log i log i η
2.3RT
where α is the transfer coefficient and is determined by the shape of the energy barrier that
must be overcome. The plus sign stands for anodic reactions and the minus sign for cathodic
reactions. i0 is the exchange current density, that is the current density at the reversible
voltage (Erev) (at (η=0)). This potential is often called equilibrium potential and can be found
in tables in most electrochemistry related books. The Tafel equation is commonly used to
find the so-called Tafel slope (b ) and the exchange current density for comparing
.
the activity of the electrocatalyst. Obviously, a good electrocatalyst has a high exchange
current density and low Tafel slope.
Equation (13) shows that the activation overpotential increases logarithmically with the
current density (A/cm2). Therefore, if a catalyst has a large actual surface area, is rough and
porous, the less current is actually applied to each location resulting in lower activation
energy. If, for example, 1 A current is applied between anode and cathode in an electrolysis
cell with highly polished electrodes having visual surface area of 25 cm2, the actual current
density on the electrodes will be 0.04 A/cm2. If however the same current is applied to a
surfaces with the same visual surface area but much larger actual surface area, for example
with roughness factor of 1000. The actual current density becomes 1000 times smaller than
for the previous case or 4e-5 A/cm2. In that way the electrolysis cell can be operated with low
overpotentials at much higher current densities than before.
The hydrogen reaction in an alkaline media is widely accepted to be a combination of the

Volmer-Heyrovsky-Tafel mechanism [18]. The activation energy for the mechanism
depends of the rate of each step. The slowest step is therefore called the rate determining
step. The first discharge step where hydrogen is adsorbed at the electrode surface is known
as the Volmer reaction and the second step where hydrogen molecules are formed is known
as the Heyrovsky reaction. The recombination of two adsorbed hydrogen atoms is known as
Tafel reactions. Both Volmer and Heyrovsky reactions are electrochemical reactions whereas
the Tafel reaction is solely a chemical reaction. In Fig. 5 the reaction steps for the hydrogen
evolution reaction (HER) as they take place in alkaline solutions are schematically
illustrated.
Fig. 5: Elementary reaction steps of HER in alkaline media.
Due to the high population of reactants, that is water, HER does typically not have any
diffusion limitations [18]. Most theories state that the adsorbed hydrogen atoms combine into
hydrogen molecules either by reacting with further discharging H2O or by recombining with
another adsorbed hydrogen atom. That is either by Volmer –Heyrovsky or Volmer – Tafel
pathways. The two discharging steps occur simultaneously and the slower step determines
the HER rate.
From the previous it is clear that the ability of a catalyst to adsorb hydrogen atoms plays a
key role in the mechanism and kinetics of hydrogen electrodes [19]. The ability to absorb
hydrogen depends on the ability of the surface to bond with hydrogen, i.e. the metal
hydrogen bond strength (H-M). First of all the absorbed hydrogen changes the free energy of
the Volmer step by the amount equal to the free energy of formation of the M-H bond. This
means that the Volmer reaction can occur at potential that is -Gads more positive to the
equilibrium potential of the reaction [19]. Secondly, the absorbed hydrogen makes the two
other reaction steps possible.
The activation energy of HER (ηact) obviously decreases with increased adsorption energy
(M–H bond strength), while increased adsorption energy means increase in terms of Hads
coverage on the electrode surface. Therefore if the M-H bond energy is too strong for the
Tafel reaction to take place, the Hads will occupy the available surface sites and inhibit the
second step of the total reaction. Thus the best hydrogen electrode should be the one having
an intermediate M–H bond energy (or free energy of hydrogen adsorption (-Gads)), as stated
in the Sabatier principle [36]. When plotting the electrocatalytic activity (exchange current
density for HER) versus the M–H bond strength for different metals, a so-called volcano plot
is formed. The volcano plot in Fig. 6 supports Sabatier’s theory and shows clearly that
platinum should be the most active metal for hydrogen evolution.
144
Fig. 6: The deependence of th

he electrocatalyytic activity for HER on the meetal – hydrogenn bond formed [18]
A more modern way to define the volcano cu urve principle is based on the elecctronic
coonfiguration of the atom t catalyst material [200]–[23] or th
ms within thee lattice of the he so-
caalled hypo- hyper-d
h theorry. Here mettals on the left side of thee volcano ploot are called hypo-
d--electronic metals
m becausse they havee empty or half-filled
h vacant d-orbitaals and the metals
m
onn the right sidde of the vollcano plot is called hyperr-d- electronic elements bbecause they y have
innternally pareed d-electron
ns which are not availablee for bonding g in pure mettals. The opttimum
(bbest catalyst, the catalystt at the top oof the volcanno plot) is either definedd to be at d8 or d5
[224]. It has beeen shown th hat by combiining hypo-d d-electronic metals
m with hhyper-d-elecctronic
m
metals synergeetic electrocaatalytic effecct resulting in
n higher actiivity for HER R is reached [24]–
[229] NiMo beiing the most popular com mposition.
As mentionedd before the actual

a electroocatalytic efffect does no
ot only depennd on the eleectron
coonfiguration. The structure and topoggraphy of thee catalyst alsso has a greaat influence on o the
appparent elecctrocatalytic efficiency and many authors hav ve publishedd an increaase in
eleectrocatalytiic activity tow
wards the HE ER by selecttively leachinng one or moore elementss from
m
metal alloys [28],
[ [30]–[3
37]. In addittion the rate of the reacttion is affect
cted by the crystal
c
orrientation and the interacction betweeen the neighb bouring speccies. Also, laattice defectss such
ass dislocationss, kinks, vaccancies and sstacking faullts are often recognised aas active siddes for
thhe HER [38].
Thhe mechanism m for the oxygen evolutiion reaction (OER) is mo

ore complex than the hyd
drogen
evvolution reaction. Severral pathwayys have beeen suggested d were the most comm monly
acccepted for alkaline waterr electrolysiss are [10]:
O
OH → OH
H e (14)
OH OH →O H O e (15)
O O →O (16)
Where one of the charge transfer steps is the rate determining step. The overpotential of the
most common OER electrocatalysts is listed in Table 4 in section 3.1.1.2.
The rate of an electrocatalytic reaction, such as the HER, depends not only on the activity of
the electrocatalyst. The rate is also determined by the composition of the electrolyte closest
to the catalyst. The area is called the double layer and is not taken into consideration in the
derivation of the Tafel equation [39]. The electrocatalytic double layer is found on the
transition zone where the charge transport changes from being an electron transport to ion
transport [40]. The density of electrons closest to the electrode surface depends on the
potential of the electrode. In the case of the cathode for HER this means that the amount of
OH- and K+ closest to the electrolyte is dependent of the electrode potential. Therefore,
relatively to the electrolyte the metal layer is seen as negatively or positively charged
depending of which ion is dominant. The double layer acts as a capacitor to the electrode
reaction due to the charging and discharging of the electrode layer closest to the electrolyte
and the change in the electron density of the metal phase. Hence, the double layer
phenomenon should therefore be taken into account for kinetic evaluations of
electrocatalysts. Thorough explanation of the double layer phenomenon can be found in
various textbooks such as [19], [41]–[43].
16
3 Water Electrolysis Technologies

The water electrolysis technique is not at all a new invention. The technique was developed
over two centuries ago by the two companions Nicholson and Carlisle [44]. Today there are
three main types of water electrolysis technologies available; these are alkaline electrolysis,
polymer electrolyte membrane (PEM) electrolysis (also named solid polymer electrolysis
(SPE)) and solid oxide electrolysis cell (SOEC). Alkaline and PEM electrolysis have both
reached commercialisation while SOECs are still in the development stage. Commercial
alkaline and PEM electrolysers are typically operated at temperatures below 100°C while
SOECs are operated at gas phase conditions at temperatures in the range of 800-1,000°C [6].
The advantage of the high operation temperature is a significant reduction in the electrical
energy demand for hydrogen production. Consequently, developers of water electrolysis
systems are increasingly looking in the direction of higher operation temperatures. The
drawbacks of working at elevated temperatures are however greater challenges regarding the
decomposition of materials, which often causes corrosion of metals and the degradation of
polymers used for sealing and etc., which in fact are the main reasons for the SOECs still
being at the R&D stage.
For the commercially available electrolysers, PEM electrolysers display the best efficiencies
at higher current densities. The investment cost for PEMs is, nevertheless, at least 10 times
larger than for their alkaline counterparties. The capital cost for commercial electrolysers are
estimated to be less than $1,000/kW for the largest alkaline systems compared to more than
$10,000/kW for small PEM electrolysers [16]. Additionally, the durability of the materials
for PEM is much less than for alkaline electrolysis. The lifetime of commercial alkaline
electrolysers is said to be about 100,000 hours compared to 10-50,000 hours for PEM
electrolysers [45]. The performance data for the main commercial PEM and alkaline
electrolysers are listed in Table 1.
Water Electrolysis Technologies 17
Manufacturer Technology Production Power Energy Efficiencya Max, H2 Location

rate (kW) consumption pressure purity
(Nm3/h) (kWh/Nm3) (bar) (vol.%)
AccaGen Alkaline 1-100 6.7- 6.7-4.87b 52.8-72.7 10 99.9 Switzerland
(bipolar) 487 (optional (99.999d)
30)
(200c)
Avalence Alkaline 0.4-4.6 2-25 5.43-5 65.2-70.8 448 n.a. USA
(monopolar) (139 c) (750c)
Cland Alkaline 0.5-30 n.a. n.a. n.a. 15 99.7
(bipolar) (99.999d)
ELT Alkaline 3-330 13.8- 4.6-4.3 76.9-82.3 Atm. 99.8- Germany
(bipolar) 1518 99.9
ELT Alkaline 100-760 465- 4.65-4.3 76.1-82.3 30 99.8- Germany
(bipolar) 3584 99.9
Erredue Alkaline 0.6-21.3 3.6- 6-5.1b 59-69.8 2.5-4 99.3- Italy
(bipolar) 108 99.8
(99.999d)
Giner PEM 3.7 20 5.4e 65.5 85 n.a. USA
(bipolar)
Hydrogen Alkaline 10-500 43- 4.3 82.3 Atm. 99.9 Norway
Technologies, (bipolar) 2150
division of Statoil
Hydrogenics Alkaline 10-60 54- 5.4-5.2b 65.5-68.1 10 99.9 Canada
(bipolar) 312 (optional (99.999d)
25)
Hydrogenics PEM 1 7.2 7.2b 49.2 7.9 99.99 Canada
(bipolar)
H2 Logic Alkaline 0.66- 3.6- 5.45-5b 64.9-70.8 4 99.3- Denmark
(bipolar) 42.62 213 (optional 99.8
12) (99.999d)
Idroenergy Alkaline 0.4-80 3-377 7.5-4.71 47.2-75.2 1.8-8 99.5 Italy
(bipolar)
Industrie Alkaline 110-760 511.5- 4.65-4.3 76.1-82.3 32 99.8- Switzerland
HauteTechnologie (bipolar) 3534 99.9
Linde Alkaline 5-250 n.a. n.a. n.a. 25 99.9 Germany
(bipolar) (99.998d)
PIEL, division of Alkaline 0.4-16 2.8-80 7.5b 50.6-70.8 1.8-18 99.5 Italy
ILT Technology (bipolar)
Proton OnSite PEM 0.265-30 1.8- 7.3-5.8 48.5-61 13.8-15 99.999 USA
(bipolar) 174 (optional
30)
Sagim Alkaline 1-5 5-25 5 70.8 10 99.999 France
(monopolar)
Teledyne Energy Alkaline 2.8-56 n.a. n.a. n.a. 10 99.999 USA
(bipolar)
System Treadwell PEM 1.2-10.2 n.a. n.a. n.a. 75.7 n.a. USA
Corporation (bipolar)
Table 1: An overview of the main current electrolyser manufacturers, their product and performance data. The
table is adopted from [46].
n.a: information not available

a
Calculated according to the HHV of hydrogen (3.54 kWh/Nm3)
b
Based on the global hydrogen production system
c
In development
d
With an additional purification system
e
Only based on the electrolysis process
As seen from the table, alkaline electrolysers have by far the largest power capacity and are
available up to the 3.5 MW range, compared to the maximum of 175 kW for PEM. The low
power consumption and, thus, low production rate for PEM electrolysers makes alkaline
electrolysis the current standard for large-scale hydrogen production among water
decomposition techniques.
18
In the following chapter the three main types of water electrolysis technologies, alkaline,
PEM and SOEC, will be introduced. Due to the nature of the thesis, a more detailed
overview will be fashioned for the alkaline electrolysis technique compared to the other two.
3.1 Alkaline Water Electrolysis
Alkaline water electrolysis is the most mature water electrolysis technology and already in
1920s several MW plants had been produced worldwide [47]. The technique is considered
to be simple and durable and lifetime up to 90-100.000 operation hours have been reported
[45],[48]. Typical efficiency for commercial alkaline electrolysers is 60-75% and 80-85% for
the best small scale systems [7]. Commercial alkaline electrolysers are typically operated in
a liquid electrolyte containing 25-40% potassium hydroxide at a temperature ranging from
60-90°C [7],[10],[49]. The largest cell efficiency losses for alkaline electrolysis originates
from the activation energies for hydrogen and oxygen, gas bubbles in the electrolyte and gas
bubble coverage on the electrode surfaces. All of these overpotentials are related to the
current density. Accordingly, the ohmic drop in an electrolysis cell increases dramatically
with increased current density. Hence, in order to maintain moderate efficiencies of up to
82% HHV, the current density has to be kept relatively low or in the range of 100-400
mA/cm2 [16]. The electrolysers are typically operated at 1-30 bar, depending on their
application. High pressure operation can reduce the ionic resistance caused by gas bubbles in
the electrolyte, due to shrinkages of bubbles, and save the cost of compressing hydrogen
after production.
The electrolysis cell consists of two electrodes; an anode and a cathode, separated by an
ionic conducting diaphragm. The diaphragm further serves as a gas separator to prevent
mixing of hydrogen and oxygen gases during operation. The operational principle for AWE
is illustrated in Fig. 2 (in chapter 2). When current is applied between the two electrodes,
water molecules surrounding the cathode are decomposed into hydrogen (H2) and hydroxyl
ions (OH-). The hydroxyl ions, which are negatively charged, migrate through the
diaphragm to the positively charged anode, where water and oxygen are formed. The half-
cell reactions and the charge carriers for the three main water electrolysis processes are listed
in Table 2.
Technology Cathode (HER) Anode (OER) Charge carrier

Alkaline H2O +2e-  H2+2OH- 2OH-  O2+H2O+2e- OH-
PEM 2H++2e-  H2 H2O  O2+2H++2e- H+
SOEC H2O+2e-  H2+O2- O2-  O2+2e- O2-
Table 2: Half-cell reactions and charge carriers for the three main types of water electrolysis [49]
3.1.1 Cell components
3.1.1.1 The diaphragm
The purpose of the diaphragm is to keep the produced gases in each cell compartment to
prevent recombination and contamination.
The criteria for the diaphragms are:
 Permeable for hydroxide ions and water

 Impermeable for gases
 Mechanical and chemical resistance to the electrolysis media
 Low ohmic resistance
Diaphragms typically have higher ionic resistivity compared to the electrolyte. In order to
minimize the potential drop over the diaphragm, the diaphragms are produced as thin as
possible. There is, nonetheless, a trade-off between reduction of thickness of the diaphragm
and its mechanical stability.
Earlier the diaphragms for alkaline water electrolysers were made of asbestos. Now asbestos
is prohibited due to its toxicity. Composite materials based on micro porous polymers or
ceramics such as polyphenylene sulfide (Ryton®) [50] and polysulfone bonded
ZrO2 (Zirfon®) [51], have gradually substituted asbestos in the newer generations of alkaline
electrolysers [52]. Scanning electron micrographs of a Zirfon® Perl 500 UTP diaphragm
applied in this study is shown in Fig. 7. The white particles in the figure are ZrO2 powder
and the grey mesh is the polysulfone matrix.
Fig. 7: Scanning electron micrograph of a Zirfon® Perl 500 UTP diaphragm utilised in this study. The white
particles are ZrO2 powder and the grey mesh is polysulfone.
3.1.1.2 The electrodes
The two electrodes, anode and cathode, at each side of the diaphragm must to be stable in the
electrochemical cell, i.e. they should not corrode, and at the same time be a good catalyst for
the electrochemical reaction taking place at their surfaces. Platinum is stable in alkaline
environment and is known to be the best electrocatalyst for water electrolysis, especially for
the hydrogen formation [18]. However, due to its high price, other less expensive materials
have replaced platinum electrodes in AWE systems.
20
Among un-noble metals, nickel is one of the most stable in strong alkaline solutions [53],
[54]. Nickel is also a relatively good catalyst for hydrogen and oxygen formation. Nickel or
nickel plated substrates are therefore typically the core material used in electrodes for AWE
systems [6],[10][45]. The nickel electrodes are typically activated by adding sulphur to the
coatings or by producing a so-called Raney nickel structure on the surface. The method of
sulphur activation of nickel electrodes dates back to at least 1923 with a Germanic patent
thereof [55]. More recently, in 1978, Norsk Hydro (now NEL Hydrogen) patented a
sulphurising process presented to give more mechanically stable cathodes than the former
patent [56]. The effect of adding sulphur to the nickel coating is not fully known. It has been
suggested that it is the formation of strongly absorbed hydrogen in the Ni-S structure being
the reason for the increase in catalytic efficiency, compared to pure nickel electrodes [57].
In 1961 Justi and Winsel [58] discovered that Raney nickel (originally developed as a
catalyst for hydrogenation of vegetable oils) was an effective hydrogen electrocatalyst for
alkaline electrolysis. The principle behind Raney nickel catalysts is that aluminium or zinc is
selectively leached from a NiAl or NiZn alloy. Lattice vacancies formed when leaching
result in a large surface area and a high density of lattice defects, which are active sites for
the electrocatalytic reaction to take place [38]. Since then, increasing the surface area and
altering the electrocatalytic configuration of an electrocatalyst by selectively leaching one or
more element from the metal alloys has widely been used to promote the activity of
hydrogen electrocatalysts [15],[28], [32], [34], [37], [59], [60]- [63].
Beside the Raney nickel and nickel sulphide activation, attempts have been made to increase
the electrocatalytic performance of nickel cathodes by doping with active substances such as
Fe, Co and Mo [59],[63]–[66]. The stability of the dopants during operation is, however,
questionable and the deactivation of Ni-Mo electrocatalyst over time has been reported [66],
[67]. An overview of the hydrogen overpotential for different hydrogen electrocatalysts
found in the literature is shown in Table 3.
Mixed oxides such as LaNiO3, NiCo2O4 and Co3O4 as well as Raney nickel and Raney Co all
display a high activity as oxygen electrocatalysts for AWE [68]-[75]. The literature implies
that despite of the large amount of work directed towards finding the optimal oxygen
electrode, few of the developed electrocatalysts have the sufficient durability or a low
enough price to be feasible for industrial electrolysers [6]. An overview of the oxygen
overpotential for different oxygen electrocatalysts found in the literature is shown in Table 4.
Composition Preparation Working Electrolyte Current ηhydrogen Publication

method temp. density (mV) year and
(C) (mA cm-2) [ref.]
Ni–Fe–Mo–Zn Co-deposition 80 6M KOH 135 83 2004 [59]
Ni–S–Co Electro-deposition 80 28 wt.% 150 70 2003 [76]
NaOH
Ni50%–Zn Electro-deposition N/A 6.25 M 100 168 2002 [60]
NaOH
MnNi3.6Co0.75 Arc melting 70 30 wt.% 100 39 2000 [61]
Mn0.4Al0.27 KOH
Ti2Ni Arc melting 70 30 wt.% 100 16 1998 [77]
KOH
Ni50%Al Melting 25 1 M NaOH 100 114 1993 [62]
Ni75%Mo25% Co-deposition 80 6 M KOH 300 185 1993 [63]
Ni80%Fe18% Co-deposition 80 6 M KOH 300 270 1993 [63]
Ni73%W25% Co-deposition 80 6 M KOH 300 280 1993 [63]
Ni60%Zn40% Co-deposition 80 6 M KOH 300 225 1993 [63]
Ni90%Cr10% Co-deposition 80 6 M KOH 300 445 1993 [63]
Raney-nickel Plasma-spray of 160 30 wt.% 1000 150 1987 [15]
Al/Ni alloy KOH*
NiS-PTFE Emulsion-paint of 90 30 wt.% 1000 100 1987 [15]
NiCoS-PTFE sulphide and PTFE, KOH*
annealing at 300C
Sinter-nickel Sintering of 120 30 wt.% 1000 160 1987 [15]
carbonyl-nickel KOH*
NiS-reduced Cathodic deposition 120 30 wt.% 1000 150 1987 [15]
of rough Ni/NiS- KOH*
mixture activation of
deposit in situ
Raney Ni Cathodic deposition 90 30 wt.% 1000 250 1987 [15]
of Zn/Ni, KOH*
Raney Ni/Co- Cathodic deposition 90 30 wt.% 1000 150 1987 [15]
NiS Zn/Ni/Co and Ni KOH*
NiS Electro-deposition 60 1M NaOH 100 97 1984 [57]
Skeleton nickel Thermal arc spraying 25 1M NaOH 10 240 2007 [32]
Nickel-titanium Thermal arc spraying 80 1M NaOH 100 240 2007 [32][33]
Ni2Al3 Pressed and heated 25 1 M KOH 250 280 2004 [28]
Ni2Al3Mo0.306 Pressed and heated 25 1 M KOH 250 160 2004 [28]
Ni2Al5 Pressed and heated 25 1 M KOH 250 253 2004 [28]
Ni2Al5Mo0.233 Pressed and heated 25 1 M KOH 250 60 2004 [28]
NiAl3 Pressed and heated 25 1 M KOH 250 136 2004 [28]
NiAl3Mo0.306 Pressed and heated 25 1 M KOH 250 57 2004 [28]
NiAl5.95Mo0.66 Plasma sprayed 25 1 M KOH 250 67 2004 [28]
Ni-Mo Synthesised 25 2 M KOH 20 70 2013 [27]
nanopowder
Cu/Ni Electro-deposition 25 1 M KOH 100 350 2011 [78]
Cu/NiCoZn Electro-deposition 25 1 M KOH 100 140 2011 [78]
Cu/NiCoZn– Electro-deposition 25 1 M KOH 100 100 2011 [78]
Ag
Cu/NiCoZn–Pd Electro-deposition 25 1 M KOH 100 104 2011 [78]
Cu/NiCoZn–Pt Electro-deposition 25 1 M KOH 100 96 2011 [78]
C/NiMn Electro-deposition 25 1 M KOH 100 141 2013 [79]
C/NiMnZn Electro-deposition 25 1 M KOH 100 127 2013 [79]
C/NiMnZ-PtPd Electro-deposition 25 1 M KOH 100 121 2013 [79]
C/NiMnZ-PtRu Electro-deposition 25 1 M KOH 100 118 2013 [79]
Table 3: An overview of the hydrogen overpotential for some electrocatalysts for AWE found in the literature.
* The author states that the electrolyte giving the highest conductivity at the working temperature is used
(compatibly with material corrosion). It is however indicated in the text that most of the electrodes are measured
at 30 wt.% KOH.
22
Composition Preparation Working Electrolyte Current ηhydrogen Publication

method temp. density (mV) year and
(C) (mA cm-2) [ref.]
LaNiO3 co- 25 1M 100 315 1982 [80]
precipitation KOH
La0.5Sr0.5CoO3 Spray-sinter 90 50 wt.% 100 250 1989 [74]
KOH
Ni0.2Co0.8LaO3 Plasma jet 90 50 wt.% 100 270 1989 [74]
projection KOH
Raney Ni Electro- 90 50 wt.% 100 280 1989 [74]
deposition of
Ni-Zn, heat
treated f. 12 h.
@400 C
Raney-Co Spray-sinter 90 50 wt.% 100 230 1989 [74]
KOH
Co3O4 Spray -sinter 90 50 wt.% 100 240 1989 [74]
KOH
Ni-Ir Electro- 25 5M 20 270 1990 [72]
deposition KOH
Ni-Ru Electro- 25 5M 20 280 1990 [72]
deposition KOH
Li 10% doped Spray pyrolysis Room 1M 1 550 2004 [81]
Co3O4 temp. KOH
Ni + Spinel type Thermo- 25 1M 100 235±7 2007 [82]
Co3O4 decomposition KOH
Ni +La doped Thermo- 25 1M 100 224±7 2007 [82]
Co3O4 decomposition KOH
MnOx modified Au Electro- 25 0.5 10 300 2007 [83]
deposition KOH
NiCo Aerogel Epoxide 25 1M 100 184 2011 [84]
addition KOH
process
NiFe(OH)2 Electro- 80 1M 500 265 2011 [85]
deposition NaOH
Table 4: An overview of the oxygen overpotential for some electrocatalysts for AWE found in the literature.
As the present section indicates, a great deal of work has been devoted to the development of
electrodes for AWE during the past 90 years. The state-of-the-art electrodes have, however,
not changed much during the years. Among the newly developed electrocatalysts, durability
measurements are usually lacking and few of the published electrocatalysts have actually
been tested in real electrolysis systems and at current densities appropriate for them. One of
the best electrolysis cell performances measured in an electrolysis stack originate from two
different R&D hydrogen programs carried out in the 1980s and 1990s [86],[87]. In the
earlier work, carried out at the Belgian Nuclear Research Centre (S.C.K,/C.E.N.), a cell
voltage of 1.6 V measured at 90°C and 0.2 A/cm2 is reported. The electrodes are described as
perforated nickel, coated with Ni-S at the cathode and spinel oxides containing Ni and/or Co
at the anode. In the latter research program, carried out at the German Aerospace Center
(DLR), vacuum plasma sprayed electrodes were used. The cathode was made of Mo
containing Raney nickel and the anode of spinel oxides of Raney Ni/Co. They published cell
voltages ranging from 1.6 to 1.65 measured at 80°C and 0.3 A/cm2. Both of the research
programs used a zero-gap cell structure, as described in section 3.1.2.2.
As mentioned above, the literature implies that the state-of-the-art electrodes used for
industrial electrolysers today, both anode and cathode, is sulphur or Raney activated nickel
or nickel coated steel [6],[10]. By combining the development work published on
electrolysers and electrocatalysts, rough ideas about the state-of-the-art electrodes used in the
industry today can be made. The exact structure and configuration of the electrodes is,
nevertheless, kept confidential.
3.1.1.3 The electrolyte

The good conductivity, compared to other bases, and the less corrosive properties, compared
to acids, makes potassium hydroxide (KOH) the most commonly used electrolyte for water
electrolysis systems [47]. The conductivity of the KOH electrolyte depends on the
temperature and concentration. In order to minimize the electrolyte resistivity in an
electrolysis system, it is essential to find the concentration which gives the highest
conductivity at the operating temperature. In 1997 See and White presented a thorough study
of the conductivity of aqueous KOH in the temperature range of -15 to 100°C for
concentrations of 15-45 wt.% [88]. More recently, Gilliam et. al. used the available
conductivity data for KOH to develop an equation for calculating the conductivity of KOH
in the range of 0-12 M and 0-100°C [89]. Using their results, Allebrod et. al. redrew a 3D
plot of the conductivity of aqueous KOH as a function of temperature and concentration, see
Fig. 8.
Fig. 8: 3D plot of the conductivity of aqueous KOH as a function of temperature and concentration [90]. Courtesy
of Frank Allebrod.
As seen from the figure, the highest conductivity for aqueous KOH in the range of 80-100°C
is reached between 30 and 40 wt.% KOH. The pH of the electrolyte rises in proportion to
the amount of hydroxide ions in the solution according to [91]:
14 (17)
where:
(18)
24
which for a strong base such as KOH can be written as:
(19)
Increasing the amount of KOH leads to higher pH and therefore a more aggressive alkaline
environment for the components of the electrolysers. Higher temperature has an increasing
effect on the corrosion rate as well. As a result, operating with KOH concentration and
temperatures slightly lower than where the optimum can be necessary for increasing the
lifetime of the electrolysis components, especially electrodes and diaphragm.
During the past three decades researchers in the field of AWE have become more aware of
the large energy losses caused by gas bubbles in the electrolyte. Thus, more effort has been
put in analysing and finding solutions for the so-called bubble phenomena [14], [92]–[96].
Already few solutions are in place for decreasing the ohmic drop caused by the gas bubbles.
Firstly, an electrolyte flow is usually applied to endorse bubble separation from the electrode
surfaces during operation. Secondly, the electrolysis cells are pressurised in order to reduce
the gas bubble volume and, thirdly, zero gap configurations, as described in section 3.1.2.2,
are used to minimise the amount of gas bubbles in the electrolyte between electrodes and
diaphragms. These solutions do, nonetheless, only eliminate the high ohmic resistance
caused by the bubble phenomena to a limited extent [14]. More recently many authors have
reported a dramatic reduction in the ohmic drop by applying an external field such as
magnetic, ultrasonic and super gravity to the electrolysis cell [94], [97]–[100]. The external
fields promote the detachment of bubbles from the electrode surface during electrolysis. As a
consequence the current density is increased since more active sites are available for the
processes to proceed.
Another relatively new field in AWE is to apply ionic activators to the electrolyte to lower
the anodic and cathodic activation energies. The process is typically based on in-situ metal
deposition on the electrode surfaces, which are said to exhibit better catalytic activity than
ex-situ processes. For the hydrogen reaction these can be ethylenediamine-based metal
chloride complex ([M(en)3]Clx, M=Co, Ni, etc.) together with Na2MoO4 or Na2WO4 [66],
[101]–[103]. Nikolic et al. [101] reported energy savings of up to 15% for electrodes
activated in situ with [Co(en)3]Cl3 and Na2WO4, compared to non-activated electrodes. The
acceleration in electrochemical activity was explained by an increase in the actual surface
area of the hydrogen electrodes. Tasic et al. [104] propose that the enhanced catalytic
activity when applying Na2MoO4 and [Ni(en)3]Cl2 ionic activators in situ, is caused by the
synergetic effect of increased true surface area and improved intrinsic catalytic effect.
Compared to the amount of research work available for electrodes and diaphragms, an
extremely small amount of work is accessible in the field of electrolyte development. The
research field of electrolytes for AWE systems is, therefore, still unexplored and yet new
discoveries are to be expected. Surfactants that influence the wettability of the electrodes and
makes them more hydrophobic could, for instance, be beneficial for minimising the amount
of gas bubbles attached to the electrode surfaces during operation.
3.1.2 Cell configuration
3.1.2.1 Bipolar vs. monopolar

Until now only the components and configurations of the AWE unit cell have been
discussed. But in fact, industrial electrolysers are composed of multiple electrolysis cells
connected in parallel or in series. There are two primary electrolysis configurations
available, commonly known as monopolar and bipolar [46], see Fig. 9 and 10.
Fig. 9: Illustration of a monopolar stack configuration. S stands for the diaphragm separator [7].
Fig 10: Illustration of a bipolar stack configuration. S stands for the diaphragm separator and B for bipolar
electrode [7].
In a monopolar electrolyser each individual electrode is fed with a current from the outside
and has a single polarity (monopolar), i.e. it is either a cathode or an anode. In an electrolysis
stack the cells are connected in parallel. The voltage across the whole stack is the same as
the voltage across any individual cell irrespective to the number of electrode pairs in the tank
[16]. In a bipolar electrolyser, each electrode is an anode on the one side and a cathode on
the other (bipolar). Thus, every two neighbouring electrodes form a unit cell. The cells are
26
connected in series and current is fed only to the end electrodes in the stack. The voltage
across the whole stack of n cells is equal to n times the voltage of an individual cell [46].
Monopolar electrolysers have a simple and robust structure made of relatively inexpensive
parts. Individual cells are easily isolated for maintenance. The essential drawback of the
monopolar electrolysers is their large surface area which makes them more space requiring,
unable to operate at high temperatures because of heat losses and increases the risk for
potential drop in the cell hardware.
The bipolar electrolysers are more compact and generally more efficient which makes them
more common in industrial applications. They can work at higher current densities and at
higher pressure and temperature. This nevertheless introduces more challenging design
issues for preventing electrolyte and gas leakage between cells [10].
For both the bipolar and the monopolar construction it is important to minimise the space
between the electrodes and the diaphragm, in order to reduce the portion of cell overpotential
from the electrolyte resistance. If, however, the space between the electrodes and the
diaphragm is too small the flow of electrolyte, which determines the mass transport in the
electrolyte, can be limited.
3.1.2.2 Zero‐gap and non‐zero‐gap design

The electrodes and the diaphragm can be assembled in two ways, namely, non-zero-gap and
zero-gap configurations, see Fig. 11.
Fig. 11: Illustration of non-zero-gap and zer-gap configuration for AWE systems. Courtesy of
GreenHydrogen.dk.
For the non-zero gap structure, solid electrodes are placed a few millimetres away from the
diaphragm. The gases produced evolve from the electrodes and drift up in the space between
the electrodes and the diaphragms. This means that during electrolysis the electrolyte
between the electrode and diaphragm will be filled with gas bubbles resulting in a major
increase in the ohmic loss from the electrolyte. The zero-gap configuration was designed in
order to minimize this problem [105]. For the zero-gap structure the diaphragm and
electrodes are closely packed. The electrodes are perforated, so the electrolyte and gases can
flow away from the electrode/diaphragm interface and evolve on the “back” of the
electrodes.
3.2 High temperature alkaline water electrolysis (HTAWE)
By increasing the operation temperature for AWE the efficiency can be significantly
increased. A great deal of research and development work has been dedicated to the
development of AWE techniques at temperatures above 120C [106]–[109]. Recently
Allebrod et. al. [110] reported results from an alkaline electrolysis cell operated at
temperatures up to 250°C at 42 bar. The cell was measured to operate at only 1.5 and 1.75 V
potential, at current densities of 1 A/cm2 and 2 A/cm2 , respectively. This corresponds to 85-
99% efficiency according to the HHV, which is much larger than earlier reported for alkaline
systems at high current densities. For comparison, H. Vandenborre et. al. reported a cell
voltage of 1.6 V at 90°C and 1.5 V at 120°C for an alkaline electrolyser operated at 0.2
A/cm2 current density [86]. Increasing the current density to 1 A/cm2 resulted in a cell
voltage of 1.9 V and 1.8 V for 90 and 120°C respectively. However, it is important to
remember that the energy needed in order to heat and pressurise the cell is not taken into
account for the efficiency measurements made by Allebrod et.al. The developers for such
techniques usually assume availability of low cost heat sources when and if used in practise.
As for the SOECs, the HTAWEs are still in the R&D stage, where the drawbacks in the
development of the HTAWE are mainly the low stability of the materials at the elevated
temperatures. The big question is, therefore, if the efficiency benefit made by the high
temperature is sufficient to compensate for the research, development and material cost
associated with the HTAWE systems. The author is not aware of any published data for long
time durability measurements of such systems. This indicates that the R&D is still at an early
stage.
288
3..3 Polym
mer Electro
olyte Mem
mbrane (PE
EM) electrrolysis
PE EM electrolyysis is constrructed in a siimilar way as

a the alkalin
ne electrolyssis. Fig. 12. shows
s
thhe outline of a PEM electrrolysis cell.
Fig. 12: Outline

O of the opperating princip
ple of a PEM ellectrolysis cell.
Thhe main diff fference betwween the AW WE and the PEM electrrolysis is thaat an acidic solid
m
membrane, typpically Nafio on, is used aas an alternaative to the alkaline
a liquiid electrolytee. The
m
membrane alsoo acts as a gaas separator for the produ uction. Wheen current is applied acro oss the
ceell the deionnised water present
p at thhe anode deccomposes into oxygen aand hydrogen n ions
(pprotons). Duee to the presence of thee sulfonic accid (-SO3H) groups in tthe membran ne the
prrotons formed at the anod de are able too migrate to the cathode and form hyydrogen [111]. See
haalf-cell reactiions in Tablee 2 in sectionn 3.1.
PE EM electrolyysers are opeerated under similar condditions as alk

kaline electro
rolysers. Howwever,
inn contrast to the alkalinee system, PEEM electrolyysis can operrate at curreent densities up to
2
2.000 mA/cm without diminishing thhe cell efficiiency extenssively [48]. T This is due to the
tigghtly packedd structure of the elecctrodes and the membrrane, often called mem mbrane
eleectrode asseembly (MEA A), and thee thin (< 0..2 mm) and d highly connductive po olymer
m
membrane thaat assures paarticularly loow cell resisstance. PEM M electrolyseers are most often
prroduced accoording to the bipolar cell concept for proper
p uation of the gases.
evacu
Thhe acidic envvironment makes

m the deevelopment of
o non-noble metal catalyyst for the system
exxtremely chaallenging. Thus,
T the el ectrodes typ
pically consist of noblee metals su uch as
pllatinum or iridium, owing their sharee in the high price of thee electrolyserrs. Currently
y PEM
eleectrolysers only
o exist in small scaless with maxim
mum hydrog gen productiv N 3/h.
ivity of 30 Nm
Thhe small annd compact structure oof the electrrolysers enaables the ellectrolysers to be
prressurised easily [16].
W
Water Electrolysiis Technologiess 29
3..4 Solid Oxide

O Elecctrolysis C
Cell (SOEC
C)
As the name succests, thee electrolyte for SOECs is made of a solid ceraamic materiaal that
coonducts oxyggen ions. HighH operatinng temperatu ures are neceessary for atttaining accepptable
ionic conductiivity in the ceramic mem mbrane. Equaally to the PE EM, meanwhhile being thee ionic
coonductor for the system, the membraane also actss as a gas sep parator. As mmentioned before,
b
SOOECs operatte at the vapour-phase, inn the range of o 800-1000C°, allowingg a greater portion
off the requireed energy to o come from m heat instead of electricity [6]. Thhe high opeerating
temmperature caalls for the use of expenssive materialss and fabricaation methodds. The electrrolysis
ceell typically consists
c of ann electrolytee of yttria staabilized zirco
onia (YSZ), a cathode maade of
niickel and YS SZ containinng cermet annd an anode composite consistingc off strontium-doped
YSZ and perrovskites succh as LaMnnO3, LaFo3, or LaCoO3 [112], [1133]. The opeerating
prrinciples for SOECs are schematically
s y illustrated in Fig. 13.
Fig. 13: Outline oof the operating

g principle of a SOEC.
Duuring operattion water steeam is fed too the cathodee, where watter is reducedd to hydrogeen gas
(HH2) and the oxide
o ion (OO2-). The anioons migrate through the porous solidd oxide lead ding to
foormation of oxygen
o gas and release of electronss at the anodde. The half-f-cell reaction
ns are
shhown in Tablle 2 in section 3.1.
Thhe SEOC tecchnology waas adopted frrom the solid d oxide fuel cell c (SOFC) technology in the
19980s [114], yet
y the solid oxide electroolysers are sttill in the ressearch and deevelopment phase.
p
W
When disregarrding the eneergy needed to heat up th he electrolyssis cell its eff
fficiency can reach
900%, but wheen including the efficienccy loss from m low price heating,
h the overall efficciency
caan only reachh 60% [10].
30
4 Methodology for Testing and

Characterisation
4.1 Structure and morphology
Numerous electrode surfaces have been developed and manufactured in the present PhD
study. The structure and morphology of the produced surfaces were investigated by the
means of light optical microscope (LOM) and scanning electron microscope (SEM). For the
purpose of fast initial screening of the developed structures LOM or low magnification
SEMs were used. Selected specimens were investigated in high magnification SEMs where
further local microstructure investigations such as electron backscatter diffraction (EBSD)
and ion channelling contrast imaging (ICCI) could be performed.
The light optical microscope used was an Olympus GX41 with an ALTRA 20 soft imaging
system camera attached. TM 3000 Tabletop scanning electron microscope from Hitachi, with
an integrated energy-dispersive X-ray spectroscopy (EDS), or JEOL JSM 5900 scanning
electron microscope, with and integrated energy-dispersive X-ray spectroscopy from Oxford
Instruments, were used for all low resolution SEM investigations. For high resolution (HR)
SEM investigations a FEI Quanta 200 ESEM FEG and FEI Helios EBS3 were utilised. For
EBSD and ICCI investigations a FEI Helios EBS3 was used.
For cross section investigation the specimens were prepared as follows:
 Hot-mounting in resin ( more delicate specimens where cold mounted in epoxy via
vacuum impregnation)
 Grinding down to 4000 grit
 Polished with 3 µm and 1 µm diamond
 Selected specimens were mechanical/chemical polished with 0.04 μm colloidal silica
for increasing the contrast between grains
 The mounted specimens were carbon coated prior to the SEM investigations
For EBSD and ICCI investigations the specimens were prepared as follows:
 Mounted in a custom made sample holder where the specimen could be demounted
after preparation
 Grinding down to 4000 grit
 3, 1 and 0.25 µm diamond polishing
 Mechanical/chemical polishing with 0.04 μm colloidal silica
Methodology for Testing and Characterisation 31
4.2 Composition and phase analysis
The EDS instruments attached to the SEMs used were utilised for elemental analyses
of the specimens. Phase analyses base on crystallography were performed via X-ray
diffraction (XRD) on a Bruker AXS, D8-Discover instrument with Cu Kα radiation.
4.3 Efficiency and durability measurements
In order to evaluate the electrocatalytic activity and durability of the developed electrodes,
various electrochemical measurements were performed. The following techniques were
used:
 Potentiodynamic polarisation
 Cyclic voltammetry (CV)
 Electrochemical Impedance Spectroscopy (EIS)
 Electrolysis cell measurements
 Electrolysis stack measurements
The potentiodynamic, CV, EIS and electrolysis tests were carried out in specially designed
electrochemical half-and whole cells, whereas the stack measurements were performed in a
commercially available AWE bipolar stack. The construction of the test setups are presented
in chapter 5.
4.4 Pre-electrolysis
During long term durability measurements, a degradation of the electrode activity is often
seen. Fig. 14 shows a degradation trend for a high surface nickel electrode developed in this
study. The measurements are carried out in 1 M KOH, 25C at fixed cathodic potential of -
1256 mV vs. the standard hydrogen electrode (SHE) for 93 hours. The resistances from the
electrolyte and gas evolution were not compensated. Hence, the noise is due to the formation
of hydrogen gas at the operating potential.
32
Fig. 14: Degradation of a high surface area nickel cathode during 93 hours of testing at -1256 mV fixed potential.
The reason for the deactivation is changes in the electron configuration of the electrodes
making the electrodes less active hydrogen electrocatalysts. This can be due to the formation
of nickel hydride phases inside the electrode lattice, corrosion or deactivating contaminants
on the electrode surfaces.
Several authors have described deactivation of nickel cathodes during hydrogen evolution by
hydrogen absorption and subsequent formation of nickel hydride in the metal lattice [115]–
[118]. The same authors have shown that nickel electrodes can be reactivated by adding
dissolved vanadium oxide (V2O5) to the electrolyte. Using vanadium oxide in industrial
electrolysers is, however, not desired due to its high toxicity to living organisms and the
environment.
The cause of the degradation can also be contaminants from the electrolyte that are reduced
or absorbed on the electrode surface. This is especially the case for the cathode because at
the HER potential many metals can be reduced. One could think that selecting analytically
clean chemicals, such as KOH, would eliminate this problem. Nevertheless, even the
cleanest chemicals commercially available contain some traces of other substances. For the
experiment performed in Fig. 14 analytically clean KOH from Merck KGaA was used. The
fabricant reports the following contaminants, among others, in the specifications for the
chemical.
Contaminant specie [%]

Ca < 0.001
Cu < 0.002
Fe < 0.0005
Pb < 0.0005
Zn < 0.0025
Heavy metals < 0.0005
Table 5: Contaminant residues in analytical clean KOH from Merck that could influence the electrocatalitic
activity of the electrode during testing.
To investigate if some of the impurities were absorbed or deposited on the cathode used in
the durability experiment in Fig. 14, elemental analyses were made by means of X-ray
photoelectron spectroscopy (XPS). For comparison one un-activated electrode (not been
immersed in KOH of leaching of Al) and one activated electrode that had been used for two
potentiodynamic measurements followed by conditioning at -456 mV vs. SHE for 10
minutes were examined. The conditioning of the activated electrode was performed in order
to dissolve impurities that possibly have been deposited on the electrode during polarisation
and the potential was selected to be below the potential where NiO is formed. The external
residues found on the surface of the electrodes are listed in Table 6.
El1 (Un-activated electrode) El2 (Activated, polarised and El3(Electrode used from
conditioned electrode) cathodic durability
test for 93 hours
[At.%] at -1256 mV vs. SHE)
C 12.38 3.22 3.73
Fe 0.89 10.80
K 3.56
Pb 0.24
Mg 0.60 1.92
Ti 0.66
Ca 0.25 0.81
Table 6: External residues found with XPS measurements performed on high surface nickel electrodes with three
different pre-treatments.
The measuring range for the XPS analyses is about 1-10 nm. From the results listed in Table
6 it is clear that the electrode cathodically treated for 93 hours (El3) contained much more
external residues compared to the two reference electrodes. This indicates that although
highly pure analytic KOH is used, impurities are deposited onto the electrode at cathodic
potentials.
It is interesting that 10 at.% of Fe is found on the surface of El3 compared to only 0.89 at%
for El2, even though the producer of the KOH reports only < 0.0005% content of Fe.
Deposition of Fe on the electrode surface is, on the other hand, not expected to be the main
cause of the deactivation of the electrode. Fe has low hydrogen overpotential, see volcano
plot in section 2.5, and during electrolysis the Fe deposits are often seen as dendrites, which
can increase the actual surface area of the electrodes. Fe is also believed to prevent nickel
hydride formation on nickel electrodes and, thus, result in less deactivation of hydrogen
electrodes [119]. These experiments have, however, only been reported on smooth surfaces.
One could therefore argue that if Fe is deposited on high surface area nickel some of the
34
most active sides for HER will be covered with Fe. Fe has slightly higher hydrogen
overpotential compared to nickel, therefore this could result in slight deactivation.
The findings of lead on the El3 surface are interesting due to the fact that the metal is known
to have high hydrogen overpotential, see volcano plot in section 2.5. Therefore, even a
partial coverage of the surface or a few atom layers could deactivate the electrode material.
The XPS measurements strongly indicate that contaminants in the electrolyte become
deposited on the cathode surface during electrolysis. Therefore, in order to eliminate the risk
of deactivation of the hydrogen electrodes during electrolysis or electrochemical
measurements, electrolyte cleaning (pre-electrolysis) should be performed at all times.
Unfortunately, the importance of the purity of the electrolyte is often overlooked for
electrocatalytic testing. In present work the electrolyte applied was always pre-electrolysed
prior to the experiments. This was done by the means of inserting two nickel plates to the
prepared electrolyte and applying about 2 V current between the electrodes for at least 48
hours.
4.5 The three electrode- electrochemical cell and IR-drop
Electrochemical measurements are typically performed in a three-electrode electrochemical

cell. This is also the case in the present study. A typical three-electrode electrochemical cell
is schematically illustrated in Fig. 15.
Fig. 15: Schematic illustration of a three-electrode electrochemical cell.

The principle behind the setup is to measure the potential of the working electrode with
respect to the non-polarised reference electrode while current is flowing between the
working electrode and the counter electrode. A simplified electric equivalent circuit for such
a setup is shown in Fig. 16 to the left, where CE is the counter electrode, RE the reference
electrode and WE the working electrode. RΩ stand for the ohmic drop between CE and RE
and Rp is the polarisation resistance.
Fig. 16: A simplified version of the electric equivalent circuit for the an ideal three- electrode electrochemical cell
setup, to the left, and a real three- electrode electrochemical cell setup, to the right.
However in real applications there is an ohmic drop (Ru) between the RE and the WE which
means the potential of the measured WE has an error of V=IRu, where I is the current applied
to the cell. This is often called the uncompensated potential drop. Ru can be minimised by
using a high ionic conducting electrolyte and by reducing the distance between WE and RE.
This is typically done by using a luggin capillary. Hence, if this is done the potential drop
between the RE and the WE can be neglected at low current densities. The problem of the Ru
can also be solved by measuring the resistance between RE and WE before the
electrochemical measurements are performed and compensate the measured data with the
calculated resistance.
When performing electrochemical measurements where gas evolution takes place, such as
for the HER or the oxygen evolution reaction (OER), special care must be taken. The reason
is that the gas bubbles formed during the measurement result in dramatic increase of the Ru.
Furthermore, the Ru is not a constant and depends on the amount and behaviour of the gas
evolution. In this case, the potential drop between RE and WE changes from one second to
another, depending of the placement, and amount of the gas bubbles in the electrolyte. Here
measuring Ru in the beginning of the measurement is not enough. Ideally, the Ru should be
measured prior to each and every potential measurement and subtracted from the RΩ
measured. This is actually what is done with the current interrupt technique. By this method
the potentiostat interrupts the current prior to each measurement point and measures the
voltage immediately before and after the current has been interrupted. When the current is
interrupted the resistance drops immediately by the amount of Ru, but Rp drops slowly due to
the capacitor. Therefore Rbefore-Rafter = Ru and IRu can be subtracted from the next potential
measurement point. The drastic difference between potentiodynamic measurements
performed on polished nickel without compensating for the IR drop and with current
interrupt IR compensation is shown in Fig. 17.
36
Fig. 17: Potentiodynamic measurements performed on a polished nickel with and without current interrupt IR
compensation.
Unfortunately, the current interrupt IR compensation is very sensitive to noise. Using a

Faraday cage to keep the external noise out can help diminishing the noise. What is more
important is to use a low impedance reference electrode. Reference electrodes with high
impedance can cause problems like overloads and potentiostat oscillation during
measurements [120]. This problem escalates with a potentiostat designed for high
performance and high speed. The impedance of a reference electrode is typically determined
by the resistance in the junction, which separates the filling of the reference electrode from
the electrolyte, as illustrated in Fig. 15. Slow flow of the filling solution through the junction
is essential to minimise the impedance of the reference electrode. Unfortunately, this flow is
often reduced in order not to alter the composition of the solution during measurements,
resulting in increased resistance in the electrode.
For the present study, the reference electrode used was a Hg/HgO reference electrode from
Radiometer. The electrode is filled with 1 M KOH and it has a fiber junction. The impedance
of the reference electrode was measured to be about 100 kΩ. According to the application
notes from the potentiostat manufacturer the impedance of the reference electrode applied
during measurements should be less than 1 kΩ. When applying the Hg/HgO reference
electrode during potentiodynamic measurements and current interrupt IR compensation,
noisy outputs were attained. This problem escalated when measuring on highly active
electrocatalytic surfaces.
In order to reduce the noise during measurements, a low impedance reference element, here
nickel wire, was coupled to the reference electrode and a capacitor in-between, as illustrated
in Fig. 18.
Fig. 18: Schematic illustration of the parallel connection of a low impedance nickel wire to the reference
electrode in order to reduce noise during electrochemical measurements.
The capacitor ensures that the DC potential comes from reference electrode and the AC
potential from the wire. The optimal size of the capacitor to be used was determined by trial
and error and where 10 nF capacitor was selected. In Fig. 19 two potentiometric
measurements prepared on high surface nickel electrodes were measured with the current
interrupt IR compensation technique. One measurement was prepared with the nickel wire
and another prepared without the nickel wire. Clearly, for the measurement where the nickel
wire is used considerably less noise is observed.
38
Fig. 19: Potentiodynamic measurements on a high surface are nickel electrode. One measurement is prepared
with nickel wire coupled to the reference electrode as illustrated in Fig. 18 and one is measured with a standard
three electrode cell setup.
4.6 Stability of electrodes
Although the corrosion stability of the electrodes or the electrocatalyst is one of the primary
reasons for electrode development, this is usually not mentioned in the scientific
publications. If the electrode material is not stable in the electrolyte during testing some or
all of the Faradaic current goes to the corrosion process. Accordingly, the measured potential
is not the potential for decomposition of water and cannot be used for evaluating the rate of
the HER, OER or the overall electrolysis process. Assuring that an electrode designed for
water electrolysis is stable in the experimental environment is essential for proper evaluation
of the electrocatalytic behaviour.
4.7 Final remarks
Unfortunately, there are no standards for activity measurements of electrocatalysts for water
electrolysis systems. The purity of the electrolyte, IR compensation and the quality of the
potentiostat measurements are all factors that can have a large influence on the final
experimental results. Furthermore, no general acceptance is in place for how the efficiency
should be calculated. Accordingly, comparing and evaluating published results on
electrocatalyst for water electrolysis can be challenging.
Development of the Test Setup 39
5 Development of the Test Setups

5.1 First generation test setup
No commercially available test setups for alkaline electrolysis systems are available. Thus, it
was necessary to develop an electrolysis test setup before efficiency and durability testing of
the designed electrodes could be done. In the early stage of the PhD study it was decided to
design an electrolysis setup that was suitable for various types of measurements related to
alkaline water electrolysis development. The criteria for the test setup were as follows:
 Adaptable both for whole and half-cell electrochemical measurements

(meaning that the design has space for inserting of reference electrode close
to one or both of the electrodes)
 Provides easy changing of electrodes and diaphragm
 Adjustable distances between electrodes and diaphragm
 Low electric resistance in connections and wiring to the electrodes
 Corrosion resistant in > 30% KOH and working temp of > 80°C
 The cell material must be isolating in order to avoid stray current during
measurements
As mentioned before, distances between the electrodes and diaphragm, type of diaphragm
and wiring all add extra ohmic losses to the electrolysis system. Therefore the idea with the
first generation of the electrolysis test setup was to be able to adjust these parameters in
order to optimise the setup for the future full size electrolysis stack.
The first challenge in the desig process was to find an isolating material that could resist the
highly corrosive environment at elevated temperatures. Polyether ether ketone (PEEK) is
known to be relatively resistant to alkaline environment [121] and it was easily available.
As result, all components in the electrolysis test setup were made of PEEK. The electrolysis
cell was made of two identical half-cell chambers which formed an electrolysis cell when
mounted. The cell was made of stainless steel (316) and coated with Teflon (ACCOFAL
P39) coating made by ACCOAT. Fig. 20 shows computer-aided design (CAD) drawings of
the first generation cell design, whereas images of the final product can be seen in Fig. 21.
40
Fig. 20: CAD images of the first generation of the electrolysis test setup.
Fig. 21: The first generation electrolysis cell test setup. A) As designed. B) Modified
For assuring low ohmic drop from wiring, the electrodes were designed with a tail which
reached above the electrolyte level during measurements. The tail could then be connected to
the wires of the potentiostat with a crocodile clamp. The electrolyte in the cell was heated by
placing the cell on a commercially available magnetic stirrer, connected to an electronic
thermometer, which was placed inside the electrolysis cell for temperature control.
When using the as designed cell, see Fig. 21 A, precise adjustment of the distance between
the electrodes and the diaphragm was challenging. Also the electrodes tended to tilt. Due to
the fact that the electrolyte, together with the formed gas bobbles, is the major cause of the
ohmic drop in the system, it is crucial to know the exact spacing between the electrodes in
order to be able to compare two different measurements. Hence, even a small spacing error
can lead to wrong interpretation of the measurements results.
An attempt was made to solve the spacing problem by making some minor modification to
the cell, see Fig. 21 B. In the modified cell the electrodes are pressed towards the diaphragm,
and polytetrafluoroethylene (PTFE) sheets with a known thickness are used as spacers. This
setup proved to work quite fine for short periods of time. Long-time durability testing was,
however, not possible due to leakage of the cell. Furthermore, the ACCOFAL coating was
not resistant to the electrolyte. Blisters and detachment of the coating were observed after
short time of operation.
5.2 Second generation electrolysis test setup
In the next generation electrolysis cell test setup, it was decided to focus mainly on the
electrode part. It was not found necessary to be able to move or change electrodes or
diaphragm without disassembling the cell. Two cell designs were made, one for whole-cell
measurements and another for half-cell measurements. CAD drawings of the second
generation whole-cell, electrolysis cell, are shown in Fig. 22.
Fig. 22: CAD drawings of the second generation electrolysis test cell.
One of the main criteria for the second generation of electrolysis cell was high temperature
resistance > 120C in strong alkaline environment. PTFE is one of the most resistant
polymers both for acidic and alkaline environment and is suitable for operations up to 150C
42
in an aggressive environment. Thus, all components of the cell, including the beaker, were
made of solid PTFE. Instead of having specially designed electrodes, the electrodes are
placed on a nickel back plate that serves as a current collector. The nickel back plate has two
tails, one for applying current to the cell and one for measuring potential difference between
the anode and cathode.
The whole-cell setup is a closed system with a lid and an automatic water dosage system,
making it possible to perform long-time durability tests in the cell without manual addition
of water, due to evaporation of water during operation. The cell can be heated by applying
heating mats from RS Components Ltd. The electrode holder is designed to fit electrodes
with the size of 5x5 cm2 where only 22 cm2 of that area is exposed to the electrolyte. The
purpose of the electrode holders, both in the first and second generation of the test setup, is
to mask the connection wires and back and edges of the electrodes, so only the selected
surface area of the electrodes is exposed to the electrolyte. The whole cell setup was used for
durability testing carried out using an AE-PS-8080-60-T power supply coupled to a
computer for data acquisition.
5.3 Second generation half-cell test setup
The second generation half-cell test setup was designed as a typical three electrode
electrochemical cell with all components made of pure PTFE. Due to low current capabilities
of measurement hardware, low surface area is generally an advantage for half-cell setups
and, therefore, the openings for the electrodes, which is the exposed area of the electrodes,
was selected to be 2 cm2. The opening was circular in order to minimise “edge effects”, i.e.
higher current densities at the edges compared to other part of the surfaces. To assure low
ohmic drop between the working electrode and the reference electrode a luggin capillary was
used. The counter electrode was made of a nickel. The cell was placed in a Faraday cage for
minimising external noise during measurements. Images of the half-cell test setup are shown
in Fig. 23 and Fig. 24.
Fig. 23: The half-cell measurement setup as assembled and connected to the potentiostat.
Fig. 24: The construction of the electrodes inside the PTFE beaker of the half-cell measurement setup.
The half-cell measurements were carried out with a Gamry Reference 3000
potentiostat/galvanostat coupled to a computer for data acquisition.
44
5.4 Industrial electrolysis stack
A durability test was carried out in a 17 cell bipolar, non-zero gap alkaline water electrolysis
stack manufactured by GreenHydrogen.dk. An exploded view CAD image of the electrolysis
stack is shown in Fig. 25. In Fig. 26 the stacking of the electrodes inside the electrolyser is
illustrated. The electrodes in the stack measured 270 cm2.
Fig. 25: CAD image of the construction of the electrolysis stack used for durability testing. Courtesy of
GreenHydrogen.dk.
Fig. 26: The bipolar configuration of the electrolysis stack used for durability testing. Courtesy of
GreenHydrogen.dk.
The durability test was performed in combination with a demonstration project named H2-
College. In the project, 66 houses at the campus of Århus University in Herning were
powered by hydrogen where surplus power from wind turbines was used to generate the
electricity for the electrolysis process. Images of the electrolysis system and the hydrogen
storage tank are shown in Fig. 27. Further information about the project can be found
elsewhere [122], [123].
Deevelopment of the
t Test Setup 45
Figg. 27: Left: Thee electrolyser sy

ystem used for H
H2-College. Riight: The hydro
ogen storage tannk from H2-Colllege,
46
6 Preliminary Work for the Development of

New Hydrogen Electrodes
When designing and developing a new electrocatalyst the three following criteria must be
fulfilled:
 The catalyst needs to be corrosion resistant and possess long time stability during
operation and shutdowns
 The catalyst material must possess good electrocatalytic properties (have low
activation overpotential) towards the required reaction(s)
 The electrocatalyst needs to be electronically conductive
From the volcano plot for HER, see section 2.5, nickel, cobalt and iron are observed to be the
best hydrogen catalysts among pure un-noble transition metals. Also, Co3O4, Raney-Cobalt
and Raney-Nickel have been found as promising electrodes for the OER, see section 3.1.1.2.
Accordingly, nickel, iron and cobalt are considered to be possible candidates as core material
for the electrodes to be developed.
The next step for selecting a proper material for the electrode development is to assure that
the material is stable under the harsh operating conditions in the alkaline media. When the
right core metal has been selected, a method for increasing the activity of the selected
electrode surface needs to be established. In this chapter the methodology for material
selection and structure modification for the process of developing a new hydrogen electrode
for AWE will be introduced.
6.1 Material selection
Pourbaix diagrams are commonly used in electrochemistry for identifying the

thermochemical stable phases of an aqueous electrochemical system. Pourbaix diagrams are
plotted with the pH of the electrolyte on the x-axis and the potential of the metal on the y-
axis. The potential is defined according to the standard hydrogen electrode.
Fig 28-30 show Pourbaix diagrams for iron, nickel and cobalt at 80°C and 1 atm. pressure.
Prreliminary Woork for the Dev
velopment off the New Elecctrodes 47
Fig
g 28: Pourbaix ddiagrams for co
obalt in water att 80 °C.
Fig 29: Pourbaixx diagrams for iron

i in water at 80°C.
Fig
g. 30: Pourbaix diagrams for nickel
n in water at
a 80°C.
48
As the developed electrodes need to be stable in strong alkaline media (pH >14) and at
intermediate temperature (> 80), looking at diagrams at pH 14 they give an indication of the
corrosion resistivity of the metals under AWE conditions. Between the blue and the red lines
in the diagrams water is stable. Above the red line oxygen is stable (oxygen evolution can
take place) and below the blue lines hydrogen is stable (hydrogen evolution can take place).
According to the diagrams, all of the suggested metals decompose at some point inside the
potential range of AWE systems, between approximately -1 and 1 V vs SHE. Nevertheless,
the thermodynamic calculations do not include information about the kinetics, i.e. the
corrosion speed of the metals in the particular media. Hence, a corrosion investigation where
the rate of corrosion is examined is necessary to evaluate the complete corrosion properties
of the three metals.
In the beginning of the present PhD study, numerous initial electrolysis tests were carried out
with cobalt and nickel electrodes. During this work an interesting observation was made.
When high surface area cobalt specimen was stored in water for few days severe corrosion of
the cobalt was observed, see Fig. 31.
Fig. 31: Cobalt specimen after storage in water for a few days shows blue corrosion products on the surface.
As observed in the image, blue corrosion products are formed on the cobalt specimen.
Additionally, the electrolysis experiments prepared with cobalt or cobalt containing
electrodes all resulted in degradation of the cobalt. These findings coupled with the
thermodynamic assessments indicate that cobalt is not stable in alkaline environment at zero
potential. This means that if cobalt is selected as core material for the electrodes, the
electrodes should not be stored in an electrolysis stack during shutdowns. This is obviously
not feasible when working with full size electrolysis stacks. Cobalt was for that reason
rejected as a core material for the electrode to be developed.
For the purpose of selecting the right material for the electrode development, corrosion
measurements of iron and nickel where carried out in 30 wt.% KOH at 150C and 5 bar
hydrogen and oxygen pressure. The relatively high temperature was selected to accelerate
the possible corrosion mechanisms. The corrosion experiments were twofold; one where
oxygen gas was bubbled through the electrolyte and another where the gas was hydrogen.
Five specimens of each metal were placed in the corrosion measurement tanks. The tanks
were opened regularly and the test specimens weighed. The average weight change of the
tested specimens during time of exposure is plotted in Fig. 32 and 33. Negative weight loss
indicates weight gain.
Preliminary Work for the Development of the New Electrodes 49
Fig. 32: Accumulated average weight loss of nickel and iron in 150°C 30 wt.% KOH at 5 bar O2 pressure.
Fig. 33: Accumulated average weight loss of nickel and iron in 150°C 30 wt.% KOH at 5 bar H2 pressure.
As seen from the plots, the weight of nickel does not change much during time, neither in the
oxygen nor in the hydrogen tank. This is in contrast with the iron specimens where weight
loss is observed during the first weeks of exposure in the oxygen tank, indicating material
loss due to corrosion. After approximately 5 weeks of exposure in the oxygen tank the iron
specimens start to gain weight. The weight gain originates from corrosion products formed
on the iron surface during exposure. In the hydrogen tank, the iron specimens experience a
weight gain only, indicating formation of corrosion products.
50
In Fig. 34 and Fig. 35 images of the test specimens before and after exposure are compared.
For the iron specimens exposed to oxygen severe corrosion is observed. The nickel specimen
seems, however, not to corrode in the oxygen environment and the weight gain apparently
occurs due to formation of nickel oxide. For the hydrogen exposure the nickel specimen look
identical to untreated specimen. This is not the case for the iron, where corrosion products
are observed.
Fe Ni
Untreated 35 Weeks Untreated 35 Weeks

of exposure of exposure
Fig. 34: Images of iron (to the left) and nickel (to the right) untreated and after 35 weeks of exposure in 150°C 30
wt% KOH at 5 bar O2.
Fe Ni
Untreated 22 Weeks Untreated 22 Weeks

of exposure of exposure
Fig. 35: Images of iron (to the left) and nickel (to the right) untreated and after 22 weeks of exposure in 150°C 30
wt% KOH at 5 bar H2
The corrosion measurements strongly indicate that iron is not a good candidate as a core
electrode material for AWE. The results show that nickel is the most stable metal in strong
alkaline media, among active transition metals towards HER and OER. The fact that nickel
is the state-of-the-art electrode material in commercial electrolysis systems supports these
findings. For the aim of producing one solid electrode suitable for bipolar stack structure the
material selected needs to be stable both in hydrogen and oxygen evolution environment as
well as at zero potential. For that purpose nickel is selected as the core material for the
development of the new electrode.
6.2 Structural modifications
It has been mentioned various times before that the electrochemical activity of a catalyst
does not only depend on the intrinsic properties of the catalyst. The structure and geometry
of a catalyst also have an effect. Platinum black (or platinized platinum) is a good example
of this. Platinum black is known to be the ultimately best hydrogen catalyst and has been
measured to have 0.5 mV HER overpotential in 2 N sulphuric acid, whereas shiny platinum
electrode is observed to have 150mV hydrogen overpotential [124]. The reason for the large
difference in electrocatalytic activity is that platinum black has significantly larger surface
area and additional surface defects compared to shiny platinum. Although, due to its price,
platinum was not considered to be a candidate for the new electrolysis electrode its
extremely good electrocatalytic properties was an inspiration for the development. In order
to learn from the HER “master”, platinum black surface was produced and its micro-
structure inspected. The production method together with more thorough results and
discussions of the platinum black electrode can be found in Appended paper III.
Fig. 36 shows the as plated platinum black surface. The platinum surface appears to be black
(Black body) due to the high absorption of all incident electromagnetic radiation, regardless
of frequency or angle of incidence.
Fig. 36: The as-plated platinum black surface.
Fig. 37 shows micrographs of the platinum surface captured in a high resolution SEM. The
micrographs reveal the extreme large surface area of the coating characterised with a
cauliflower structure. The large surface of the platinum black implies that the electrocatalytic
activity of a surface can be largely increased by increasing the real surface area of the
catalyst and the amount of crystal defects, where the HER is suggested to take place.
522
Fig. 37: High resolu

ution scanning eelectron microsscope images off platinum blackk surface.
Acccordingly, the
t aim was to produce a nickel surfaace with a laarge real surfface area andd large
ammount of reactive crystal defects. A ccommon metthod for prep paring porouus nickel cattalysts
is by selectiveely alkaline leaching off aluminium or zinc from a Ni-Al or a Ni-Zn alloy,
reespectively. Various
V techn
niques for prroducing the alloying surrfaces have bbeen reported
d:
 Electrrodeposition of Ni-Al powwder with Nii [125]–[129]

 Powdeer pressing of Ni-Zn powderrs [62], [130]
o Ni-Al or N
 Electrrodeposition of Ni–Zn allloys [60], [13
31], [132]
 Therm
mal spraying [28], [35], [336], [133]
W
When choosinng the right process tecchnique for the t electrodee developmeent, the inteerlayer
addhesion betw ween the porrous structurre and the su ubstrate as well
w as the ppurity of thee final
prroduct must be taken intto considerattion. The hig ghly alkalinee electrolytee, the interm
mediate
temmperature (> >80C) and the oxygen and hydrogeen gas evolu ution, all conntribute in making
m
thhe electrode media
m extrem
mely corrosivve. Proper addhesion betw ween the electtrocatalyst an nd the
suubstrate is vital
v for the overall lifettime of elecctrodes for AWE.
A If thee adhesion is not
suufficient, the electrolyte can
c penetratee in-between n the two layyers resultingg in gas evoolution
inn the interphhase, leadingg to gas eroosion corrosion. The hig ghly electroochemically active
suurface will then, partly y or entirelly, be “blow wn-off” dimminishing thhe electrocattalytic
effficiency of the
t electrodee. Surely, simmilar gas eroosion corrosiion mechanissm can take place
beetween unatttached layerrs in the eleectrode coatting itself. Therefore,
T chhemical bou unding
beetween the suubstrate and the electrocaatalytic coatiing, instead of
o interlockinng, is preferrred.
Goood electrical conductiv vity is anothher importan nt factor, whhen producinng electrocattalytic
acctive surfaces. Insulating g particles, suuch as oxidees, increase the
t electricalal resistance of the
eleectrode and,, consequently, decreasess the efficien ncy of the ellectrolysis prrocess. The brittle
prroperty of thhe oxides also increasess the risk off gas erosion n corrosion inside the porous
p
strructure. Hennce, selecting g the right pprocess technnique for prroducing thee electrocatallyst is
esssential for thhe efficiency
y and durabiliity of the dev
veloped electrocatalyst.
One of the most common ways to produce large surface nickel electrodes is by thermal
spraying (TS) of Raney nickel powders (usually 50/50 wt.% Al/Ni) onto a nickel or a steel
support [28], [30]–[34], [36], [134]. During spraying the Raney powder partly melts and a
coating consisting of number of different Al/Ni phases is formed. This is followed by
selectively leaching of aluminium from the alloy(s) and a skeletal nickel structure with a
high surface area and large amount of crystal defects is formed.
In the case of the TS process, the coatings are characterized by a heterogeneous layered
pancake-shape structure, containing voids and oxide inclusion as shown in the SEM
micrograph and the corresponding illustration in Fig. 38.
Fig. 38: Left: Cross section back scatter electron micrograph of a Raney nickel coating prepared by thermal
spraying of Al/Ni powder onto a nickel substrate. Right: Schematic illustration of a thermally sprayed structure
[42].
Because the TS structure contains no chemical bounding, between the substrate and coating,
and it has oxides incorporated in the coatings, one can argue that this is not the ideal process
technique for fabricating electrocatalytic surfaces for industrial AWE. Another drawback of
the TS technique is that the coatings are relatively thick, typically over 100 µm, and rough,
i.e. unlevelled, see Fig. 38. This type of surfaces gives problems when assembling the
electrodes in an electrolysis stack.
Powder pressing is also a process where interlocking rather than chemical bounding is
obtained. With the aim of finding a new cost efficient technology for producing
electrocatalytic coatings with a large surface area and mechanical properties that can
withstand the corrosion challenges in AWE, these process techniques were not selected for
the present study. The process techniques selected for producing high surface area nickel
electrodes are introduced in the next chapter.
54
7 Manufacturing of High Surface Area Nickel

Coatings
The aim of the work, presented in this chapter, is to find a new cost efficient technology for
producing electrocatalytic nickel coatings for industrial AWE applications. The coatings
were required to have large actual surface area and mechanical properties that can withstand
the corrosion challenges in industrial alkaline water electrolysers. The following processes
and combination of processes were screened:
 Physical vapour deposition of aluminium onto an nickel plate followed by thermo-

chemical diffusion
 Hot dip aluminising of nickel followed by thermo-chemical diffusion
 Direct thermo-chemical diffusion of aluminium and nickel sheets
 Aluminium ionic liquid electroplating on a nickel plate followed by thermo-
chemical diffusion
 Physical vapour deposition of aluminium onto electroplated sulfamate nickel
substrate followed by thermo-chemical diffusion
The best coatings attained from the screening where selectively aluminium leached in order
to facilitate a porous nickel structure.
7.1 Physical vapour deposition of aluminium onto a nickel plate
7.1.1 Introduction
The first trial for producing high surface area nickel electrocatalytic surface was by physical
vapour deposition (PVD) followed by thermo-chemical diffusion and alkaline leaching. The
PVD technique was selected because of its superior interlayer adhesion and the possibility of
high purity coatings. Moreover, the thickness of the coated layer can be controlled precisely
assuring unique uniformity. The selected PVD techniqe was of the magnetron sputtering
type, where material is ejected from a sputter target due to bombardment of ions to a
substrate surface [42]. In the particular case very pure aluminium metal are sputtered to the
electrode surface forming a thin film of aluminium.
7.1.2 Experimental procedure
Commercially available nickel plates with a thickness of 0.5 mm were used as electrode
substrate. The purity of the nickel plates were determined by optical emission spectroscopy,
detecting 99 wt.% Ni, 0.25 wt.% Mn, 0.14 wt.% Fe and 0.11 wt.% Al. Other residual
elements were determined to be below 0.1%. The nickel specimens where coated with
approximately 20 µm of aluminium by DC-magnetron sputtering using a CC800/9 SinOx
coating unit from CemeCon AG. The aluminium source was an Al 1050 alloy target, run at
Manufacturing of High Surface Area Nickel Coatings 55
750 W, the RF bias on the substrate was set to 800 W and the start pressure in the chamber
was 1mPa. The nickel substrates were cathodically degreased for two minutes prior to the
PVD process. The substrates were heated and etched in situ by Ar sputtering, prior to the
sputtering process, to remove nickel oxide (NiO) from the surface. The PVD aluminium
coated nickel specimens were subsequently heat treated in an atmospheric furnace for 24
hours at 610°C followed by a selective aluminium leaching.
Among Al/Ni alloys, aluminium can only be alkaline leached from the Al3Ni2 and Al3Ni
phases [135]. Obviously, the Al3Ni2 phase contains more nickel and is therefore more
mechanically stable compared to its Al3Ni counterpart. The aim during the alloy formation
procedure, i.e. the thermo-chemical diffusion process, was to obtain a relatively thick layer
of the Al3Ni2 phase. The 24 hours of heat treatment was inspired from the calculated
diffusion coefficients of the Ni-Al system in [136].
The first batch of the heat treated PVD Al/Ni specimens were leached as follows; 2 hours in
1 wt.% NaOH at room temperature followed by 20 hours in 10 wt.% NaOH at room
temperature and 4 hours in 30 wt.% NaOH at 100C. These specimens will be referred to as
PVD Al/Ni 1. Batch two of the aluminium deposited specimens was leached in 30 wt.%
KOH and 10% KNaC4H4O6×4H2O at 80°C with stirring for 24 hours. These specimens will
be referred to as PVD Al/Ni 2. Images of a PVD aluminium deposited nickel plate as
received, after heat treatment and after heat treatment and alkaline leaching are shown in Fig.
39.
Fig. 39: PVD aluminium deposited nickel substrate, as received, heat treated and alkaline leached.
The appearance of the leached specimen is similar to platinum black. This indicates that a
large surface area has been produced.
56
7.1.3 Results and discussions
Untreated nickel substrate and aluminium coating in the as-deposited state were examined in
SEM. The back-scatter electron (BSE) micrographs in
Fig. 40 display parallel grooves on the nickel surface from the rolling process and large
crystallites which characterise the annealed nickel substrate. The secondary electron (SE)
surface images in Fig. 41 show how the PVD aluminium coating imitates the topography of
the substrate, resulting in parallel grooves in the aluminium structure.
Fig. 40: BSE micrographs of nickel substrate, left: surface, right: cross section.
Fig. 41: SE micrograph of the PVD aluminium in as deposited state in different magnifications.
In Fig. 42 a cross section of a heat treated PVD Al/Ni electrode is shown together with the
Ni-Al binary alloy phase diagram. Results from cross section elemental analyses are shown
in Table 7.
Fig. 42: Left: Ni-Al binary alloy phase diagram from [137]. The horizontal line indicates the thermo-chemical
diffusion temperature (610C). Right: Cross section of a PVD Al/Ni electrode after heat treatment at 610°C for
24h, prior to leaching. The arrows indicate the supposed intermetallic phase found in the cross section. The
numbers refer to the EDS analysis in Table 7.
Name of Before leaching Leached PVD Al/Ni 1 Leached PVD Al/Ni 2

phase
/original
phase
Phase Al Ni O Al Ni O Al Ni
nr. [wt.%] [wt.%] [wt.%] [wt.%] [wt.%] [wt.%] [wt.%] [wt.%]
1 Ni2Al3 37 63 4 21 75 7 13 80
2 Ni2Al3 37 63 36 64 4 15 81
3 NiAl 30 70 30 70 29 71
4 Ni3Al 13 87 13 87 14 86
5 Ni 100 100 100
Table 7: Results from the cross section EDS analysis on PVD Al/Ni electrodes before and after the first and the
second leaching procedure. The phase numbers refer to the numbers in Fig. 42 and Fig. 43. All elements from the
periodic table are analysed.
Comparing the EDS data with the Ni-Al phase diagram, it is supposed that the three
following Al-Ni intermetallic phases are formed; Ni2Al3, NiAl and Ni3Al, seen from the top
towards the pure nickel substrate. This is also indicated with the green arrows in Fig. 42.
These assumptions are in agreement with the findings of Janssen and Rieck [136]. The
majority of the intermetallic phase formed is the strong, and yet leachable, Ni2Al3 phase
[135].
When heat treated at 610°C, a thermo-chemical diffusion process takes place at the interface
between the aluminium and nickel. The aluminium atoms diffuse into the nickel structure
and thermodynamically alloys can be formed. The red horizontal line in the phase diagram in
Fig. 42 indicates which Ni-Al diffusion couples are thermodynamically stable at 610°C and
atmospheric pressure. The thickness of each intermetallic phase formed depends on the
diffusion kinetics, the amount of nickel and aluminium available in the diffusion system and
the heat treatment parameters.
Fig. 43 shows cross section SEM micrographs of the PVD Al/Ni electrodes after heat
treatment followed by the first and second alkaline leaching procedure, PVD Al/Ni 1 and
PVD Al/Ni 2, respectively.
58
Fig. 43: Cross section SEM micrographs of a PVD Al/Ni electrode after heat treatment at 610°C for 24h,
followed by alkaline aluminium leaching. Left: Prepared by the first leaching procedure (PVD Al/Ni 1). Right:
Prepared by the second leaching procedure (PVD Al/Ni 2).
For the first leaching procedure, an about 5 µm skeletal Al/Ni residue is formed. When
selectively leaching the aluminium with the second leaching procedure the entire Al3Ni2
phase is leached, resulting in considerably larger thickness of the porous nickel layer. The
difference is due to the slow leaching kinetics of the Ni2Al3 phase below 70°C in alkaline
media [135]. In addition, some of the aluminium could possibly have been redeposited into
the pores during the first leaching method since no KNa-tartrate tetrahydrate was used in that
process [138]. The reason for adding KNa-tartrate tetrahydrate to the alkaline leaching
solutions is that it works as a complexing agent for the leached aluminium according to
[139]:
→ (20)
hereby preventing the aluminium hydroxide precipitates to redeposit into the nickel pores
[140]:
→ (21)
Producing high surface area nickel electrodes by PVD of aluminium onto a nickel plate was
found to be successive. Numerous efficiency, durability and metallography investigations
have been performed on the structure, see chapter 8, Appended papers I,II and III and the
Appendix. After the successful experience of the first generation of electrodes for AWE,
other process techniques for producing similar electrodes were assessed. The process
techniques were selected with the aim of reducing the production cost associated with the
electrode manufacture.
7.2 Hot dip aluminising
7.2.1 Introduction
Hot dip aluminising is a well-known process and widely used in the steel industry for
improving the wear and corrosion resistance [141]. Still, only few attempts of hot dip
Maanufacturing off High Surface Area
A Nickel Cooatings 59
aluuminising of nickel havve been repoorted [142]–[[144]. This process

p technnique was chosen
c
beecause it waas expected to give goood adhesion between the nickel ssubstrate an nd the
aluuminium coaating. The technique is, ffurthermore, inexpensive and relativeely simple.
7.2.2 Experrimental pro

ocedure
( the same type as in 7.1), measurin

Seeries of nickeel coupons (of ng approxim
mately 100×220×0.5
m
mm were imm mersed into liquid alumiinium melt (99,8% Al) at around 7700°C for 2 - 600
seeconds and suubsequently quenched inn water at room temperatture. The nicckel couponss were
immmersed in a cathodic degreaser for 2 minutes, pickled for 1 minute, rin insed with etthanol
annd dried prioor to each dippping proceddure. For miinimizing thee amount off oxide slag on
o the
sppecimen surfface after dippping, the topp layer of th
he aluminiumm melt was sscraped asidee prior
to each dippinng procedure. Fig. 44 shoows the clay crucible useed in the expperiment as placed
p
inn the inductioon furnace ap
pplied, beforee and after melting
m of thee aluminium..
Figg. 44: The clay crucible used for

f the hot dip aaluminising as placed
p in the fu
urnace before annd after melting
g of the
aluuminium.
Thhe hot dip aluminised

a nickel was suubsequently heat treated in an atmosspheric furnace at
6110°C for 24 hours.
h The long durationn of heat treaatment was selected
s to asssure formattion of
thhe Al3Ni2 phhase through hout the whoole diffusionn layer, as earlier
e proven
en successfull. The
exxperimental series
s for thee hot dip alum
minising procedure are listed in Tablee 8.
Specimenn number Dippinng-time [s] Heeat treatment aat 610 ˚C [h]

1 2 -
2 8 -
3 30 244
5 300 -
6 600 -
Taable 8: Experim
mental series forr hot dip aluminnizing
7.2.3 Resullts and discu

ussions
Thhe most com mmon challen nges when w working with h hot dip alum
minising is ddegradation of the
aluuminium meelt due to oxidation andd hydrogen dissolution [141].
[ Alumminium has a high
afffinity to oxyygen, thus, the aluminiuum reacts easily with th he oxygen iin the atmossphere
duuring meltingg and an insoluble oxidee slag is form
med on the surface of thee aluminiumm melt.
Liiquid aluminnium at 700°C C has hydroggen solubility
y of about 1 ml/100 g [1445], whereass solid
aluuminium has hydrogen solubility
s off maximum 0.034
0 ml/1000g [141]. Thherefore, hyd
drogen
600
gaas entrapped in the alum

minium melt w will result in
n voids in th
he solid alum
minium phasee. The
hyydrogen origginates from the reactionn of liquid aluminium
a an
nd moisture from the fu
urnace
atmmosphere acccording to:
2 → 3 (22)
k at 700°C, indicating th
Thhis reaction has a Gibb’ss free energyy of -787.3 kJ that the reacttion is
hiighly thermodynamicallyy favourable aat the specifiic temperaturre.
As an attemptt to reduce th he formationn of oxygen slag and hyd drogen dissollution in thee melt,
arrgon gas waas purged th hrough the lliquid alumiinium during g dipping. U
Unfortunately y, the
puurging of arggon gas into the melt cauused turbuleence and agittation in the liquid alum minium
leaading to signnificant increease of oxidaation of the melt.
m Specim men B in Fig. 45 characterrises a
sppecimen afteer hot dip alluminising w where argon n gas has beeen purged tthrough the melt,
whhereas specimen A charaacterises a hhot dip alumiinised specim men produceed without pu urging
off argon gas.
Figg. 45: Two hot dip aluminized

d nickel couponns. A) Without purging of argon gas and B) w
with purging of
o argon
gaas.
It is clear from
m the imagess that more ooxide slag is attached to the nickel cou
oupon immerrsed in
thhe liquid alumminium purgged with argoon gas. Therrefore, it wass therefore ddecided to peerform
thhe hot dip aluuminising witthout argon ppurging.
Fiig. 46 showss cross section LOM miicrographs of

o nickel plattes after bein
ing immersed into
m
molten aluminnium for 2 annd 8 seconds .
Fig. 46: Cross section LOM image of a nickel plate immersed in molten aluminium. Left: for 2 sec. Right: for 8
sec.
As seen from the LOM micrographs, 2 seconds of dipping time is not enough to achieve
proper wetting between the aluminium melt and the nickel surface. When immersing for 8
seconds, considerable better wetting of the nickel surface is attained. There is, however, still
some localised unwetted areas to be found on the surface. The inhomogeneous wetting of the
surface can be caused by different thickness of nickel oxide to be reduced by metallic
aluminium on the coupon surface or by the aluminium oxide formed during the initial
reduction of nickel oxide on the surface according to:
2 3 → 3 (23)
The small pores observed in the aluminium coating are presumably hydrogen pores from the
hot dip aluminising process.
When immersing a nickel coupon into the aluminium melt for 30 seconds the entire nickel
surface becomes wetted, Fig. 47. The aluminium layer is however, as expected from the
inhomogeneous wetting, not uniform.
62
Fig. 47: Cross section BSE SEM micrographs of a nickel coupon after hot dip aluminizing for 30 seconds.
When the nickel plate is immersed into the 700°C aluminium melt, the aluminium starts
almost immediately to diffuse into the nickel bulk and the eutectic Al3Ni phase is formed.
According to the Al-Ni binary alloy phase diagram, shown in Fig. 48, the line for the liquid
melt at 700°C is in contact with the Al3Ni phase.
Fig. 48: Al-Ni binary alloy phase diagram [137].
Therefore, some of the Al3Ni phase formed during the hot dip aluminising process will
dissolve in the aluminium melt forming Al3Ni precipitates in the aluminium rich melt. This
is evident in Fig. 47 where the white areas are Al3Ni precipitates, the grey areas the
Al+Al3Ni eutectic phase and the dark areas are aluminium. Solid state diffusion between the
Al3Ni and the nickel substrate results in the formation of a thin layer of Al3Ni2 phase. The
results from EDS analyses are shown in Table 9.
Name of phase Hot dip Al 30 sec. Hot dip Al 30 sec. + heat @ 610°C for 24h.
/original phase
O Al Ni O Al Ni
[wt.%] [wt.%] [wt.%] [wt.%] [wt.%] [wt.%]
Ni - - 100 - - 100
Al3Ni2 - * * 3 38 59
Al3Ni - 58 42 - - -
Al+Al3Ni 3 87 10 - - -
Al 3 92 5 - - -
Table 9: Results from EDS analyse on the hot dip aluminised specimen, prior to heat treatment (Fig. 47) and after
24 h. of heat treatment (Fig. 50). All elements from the periodic table except for carbon are analysed.
*The area of the intermetallic phase is too thin to be measured. The name of the phase is predicted from the Al/Ni
phase diagram.
Due to the dissolution of the nickel into the aluminium melt during the process it is
important not to immerse the specimen into the melt for excessive time periods, as this will
destroy the initial surface structure of the nickel coupons, making them very non uniform.
The cross section micrograph of a nickel coupon after 5 minutes immersion in the aluminium
melt, Fig. 49, shows the dissolution of the nickel coupon after immersion. Hot dip
aluminising a nickel coupon for 10 minutes resulted in complete dissolution of the nickel
specimen.
64
Fig. 49: Cross section LOM micrograph of a nickel coupon after 5 minutes of hot dip aluminising.
In order to facilitate larger amount of the desired Al3Ni2 intermetallic phase, a nickel coupon
that had been immersed in the aluminium melt for 30 sec. was heat treated at 610°C for 24
hours, see Fig. 50.
Fig. 50: Cross section SEM micrographs of a nickel coupon hot dip aluminised for 30 sec. and heat treated at
610°C for 24 hours.
EDS analyses of the cross section are listed in Table 9. The analyses indicate that the
developed surface solely consists of about 100 µm Al3Ni2 phase.
As apparent from the SEM micrographs, the Al3Ni2 phase formed contains cracks and voids.
The cracks are possibly due to the compressive stress induced in the coating during diffusion
and/or due to thermal mismatch stresses generated in the coating as it cools from 610°C to
room temperature. The stresses in the coating pile up and increase throughout the thickness
of the Al3Ni2 layer. The relatively brittle Al3Ni2 structure is evidently not able to
accommodate for the stresses through the whole structure. This results in lack of space for
the developed coating and cracks are formed. The voids in the developed Al3Ni2 structure are
possibly due to hydrogen and oxides that are formed in the aluminium layer during the hot
dip aluminising process as explained earlier.
The next step in the procedure for producing electrodes with large actual surface area for
AWE ought to be selective alkaline leaching of the aluminium from the Al3Ni2 phase. Due to
the brittle features of the coating, the existence of insulating oxides and the difficulties of
controlling the layer thickness of the coating, the hot dip aluminising process technique was
evaluated to be unsuitable for the AWE electrode development. Further process procedures
were, therefore, not prepared for this particular screening.
7.3 Thermo-chemical diffusion of aluminium and nickel sheets
7.3.1 Introduction
In this chapter the screening from the simplest way of diffusing aluminium and nickel
together is introduced. The process involves placement of a nickel and aluminium sheet on
top of each other and heat treated. This method was actually not expected to give good
results due to the oxide layers of the original sheets that could not be sputtered away in-situ
as for the PVD process. Due to the surprisingly good results from this trial the results will be
reported here.
Aluminium sheet (99% Al) measuring 50×50×0.5 mm was placed on top of a nickel sheet
(99% Ni, of the same type as previously) of the same size. The sheets were thereafter heat
treated in argon atmosphere furnace. The initial temperature was set to be 700°C in order to
melt the aluminium. Thereafter the temperature was gradually reduced to 610°C for
maintaining thermo-chemical diffusions and preventing nickel dissolution. The overall heat
treatment varied for 2 hours. The nickel and aluminium sheets were immersed in a cathodic
degreaser for 2 minutes and pickled for 1 minute, rinsed and dried prior to the heat treatment.
The specimens were subsequently leached in 30 wt.% KOH and 10 wt.%
KNaC4H4O6×4H2O at 80°C with stirring for 24 hours.
666
7.3.3 Resullts and discu

ussions
Siimilarly to the hot dip p aluminisinng process, Al + Al3Ni eutectic pphase is thee first
inntermetallic phase
p to be formed duriing the diffuusion processs between thhe aluminiumm and
niickel sheets. This is folloowed by a ssolid state diiffusion betw
ween the Al33Ni and the nickel
reesulting in forrmation of Al
A 3Ni2 phase according too [146]:
→ (24)
Inn contrast too the hot dip

d aluminis ing process, the direct diffusion pprocess givees the
poossibility of longer
l heat treatment
t witthout deform
ming or dissolving the nicckel substratee. The
tw
wo hours of heat
h treatmen nt results in tthe formation
n of about 10
00 µm of thee Al3Ni2 phasse, see
Fiig. 51.
Fig. 51: Cross section BSE SE

EM micrographhs of the thermo
o-chemical difffused aluminium
m and nickel sheets.
Thhe direct forrmation of AlA 3Ni2 takes place becau use the Al3Ni
N eutectic pphase is abov ve the
lower heat treaatment temperature appliied, i.e. 610°°C, see Al-Ni phase diagrram in Fig. 48.
4 As
a result, no poost heat treattment proceddure is need
ded. The attained Al3Ni2 layer is relaatively
unniform withoout any crack ks or pores. R m the EDS analyses of thee phases in Fig.
Results from F 51
arre reported inn Table 10.
Name of phase Heat treated Heat treated and Al leacheed

/original ph
hase
Al Ni O Al Ni
[wt.%
%] [wt.%] [wt.%] [w
wt.%] [wt.%
%]
Ni - 100 - - 100
Al3Ni2 43 57 13 8 79
Al3Ni 73 27 - - -
Al 100 - - - -
Taable 10: EDS annalyses on therm mo-chemical diiffused aluminiium and nickel sheets (Fig. 51 and 52). All ellements
froom the periodicc table except fo
or carbon are annalysed.
Inn order to faacilitate a po

orous nickel structure, the heat treaated specimeen was selecctively
aluuminium leaached in 30 0 wt.% KOH H and 10 wt.%w KNaC4H4O6×4H2O O. During alkaline
leaaching, the entire
e aluminnium rich topp layer of thee specimen iss etched away
ay. The majorrity of
the aluminium in the Al3Ni2 phase is selectively leached as well. This is evident from the
cross section SEM micrograph in Fig. 52 combined with the EDS analyses reported in Table
10.
Fig. 52: Cross section BSE SEM micrograph of the thermo-chemical diffused aluminium and nickel sheets after
selective leaching of aluminium.
The remaining nickel residue is characterised with large macropores perpendicular to the
surface. Smaller cracks can also been seen between the leached layer and the nickel
substrate. The cracks are formed due to the tensile stresses that build up in the structure
during aluminium leaching. The fact that the etched structure contains only 8 wt.%
aluminium and a reasonably large amount of oxygen, 13.wt%, the structure is expected to be
porous. This can however not been seen in the SEM micrographs at the possible
magnification on the applied SEM.
7.4 Aluminium ionic liquid electroplating
7.4.1 Introduction
Aqueous electroplating is the most common process technique for producing metallic
coatings with high purity and god adhesion properties. However, due to the reactive nature
of aluminium, i.e. E0 = -1.66 V vs. SHE [42], hydrogen evolution reaction will take place
prior to deposition during electroplating in aqueous media. Electroplating of aluminium can
therefore only be prepared in non-aqueous electrolytes. The two main electrolytes used for
electroplating of aluminium are non-aqueous organic solvents and molten salts. The non-
aqueous organic solvents are generally inflammable, volatile, have low conductivity and
narrow electrochemical window. Inorganic molten salts operate at temperatures above 150°C
whereas organic molten salts (i.e. ionic liquids) operate at lower temperatures [147]. Several
authors have described successful aluminium deposition using a nontoxic ionic liquid bath
containing dimethyl sulfone (DMSO2) and aluminium chloride (AlCl3) [147]–[149]. Inspired
by the literature aluminium was electroplated in an ionic liquid electroplating bath
containing dimethyl sulfone and aluminium chloride with a molar ratio of 10:2
(DMSO2/AlCl3).
68
Aluminium was electroplated in an ionic liquid electroplating bath containing dimethyl

sulfone and aluminium chloride with a molar ratio of 10:2 (DMSO2/AlCl3). The
electroplating procedure was constructed in a sealed glass beaker as illustrated in Fig. 53.
Fig. 53: Schematic illustration of the experimental setup for the aluminium ionic liquid electroplating procedure.
The operation temperature was 120°C and the deposition was prepared at the current density
of 10 A/dm2 for 2 hours in an argon atmosphere. The anode was made of 99% aluminium
and the cathode (substrate) consisted of nickel (99 wt.%). The nickel substrates measured
approximately 50×10×0.5 mm. Both anode and substrate were cathodically degreased for 2
minutes and pickled for 1 minute prior to the electroplating process. The electroplated
specimen was subsequently heat treated in an atmospheric furnace at 610°C for 24 hours.
According to [150] the plating mechanism is as follows; first blending of the solvent and the
reactive aluminium salt according to:
4 3 → 3 (25)
Thereafter reduction of aluminium according to:
3 → 3C H O S Al (26)
As seen from Fig. 54 (left) at about 100 µm dense aluminium coating is attained from the
ionic liquid electroplating process.
Fig. 54: Cross section LOM micrographs of ionic liquid electroplated Al on a nickel substrate. Left, as plated.
Right, heat treated for 24 h. at 610°C.
Fig. 55: Cross section SEM micrograph of a ionic liquid electroplated aluminium on a nickel substrate after 24
hours heat treatment at 610 °C.
The aluminium coating is characterised with relatively many pores, similar to those observed
with the hot dip aluminising process. The porous structure is possibly due to hydrogen
evolution during plating. Although the electroplating process is prepared in argon
atmosphere, the hygroscopic nature of the AlCl3 salt makes it difficult to assure that the salt
will not react with the water in the atmosphere. As said before, water is not stable at the
electroplating potential and will be reduced to hydrogen at the cathode.
EDS analyses of the as plated structure indicates that some sulphur and chlorine is
incorporated into the deposited aluminium during plating, see Table 11.
Phase Electroplated Electroplated + heat @ 610°C for 24 hours

no.
O Al Ni S Cl O Al Ni S Cl
[wt.%] [wt.%] [wt.%] [wt.%] [wt.%] [wt.%]
[wt.%] [wt.%] [wt.%] [wt.%]
1 - - 100 - 100
2 4 94 - 1 1 - 37 61
3 43 48 3 4 2
4 100
5 8 60 32
Table 11: EDS analyses on the aluminium ionic liquid electroplated nickel substrate (Fig. 54 and 55). All
elements from the periodic table except for carbon are analysed.
70
Some of the Cl and S content are expected to be from the electrolyte that gets entrapped in
the porous structure during plating. However, while electroplating a strong odour of SO2 was
detected from the plating bath, indicating degradation of the DMSO2 solvent. Jiang et al.
[150] also reported findings of Cl and S in their coatings. In their study, insufficient cleaning
of the electrodes and/or entrapped or codeposited electrolyte was assumed to be the cause of
the impurities. Recently Miyake et al. [149] reported thorough investigations of S and Cl
impurities in aluminium coatings from DMSO2/AlCl3 electrolytes at current densities
varying from 20–80 A/dm2 and molar ratio of 10:2 and10:3 (DMSO2/AlCl3). The produced
aluminium coatings were found to contain 0.1–1 at.% of Cl and S. The impurity content was
found to decrease with increased current density. Higher molar ration of AlCl3 resulted in a
slight rise in the impurity content as well. The authors did, however, not succeed in finding
the reason for incorporation of Cl and S impurities in the coating.
Here it is proposed that the S is reduced at the cathode during electrodepositing leading to
the formation of Al2S3 according to the following half-cell reaction:
3 2 12 → 3 6 3 (27)
Acetylene (C2H2) is proposed as the possible carbon compound formed by the

electrochemical reduction of dimethyl sulfone (DMSO2). The half-cell reaction has an
equilibrium potential (E0) of -0.770, which is much lower than the actual equilibrium
potential of aluminium. This means that the proposed decomposition mechanism can take
place at potentials that are less negative than the cathodic potential needed to deposit
aluminium, i.e. less energy is needed to decompose the solvent compared to electroplating of
aluminium. Al2S3 is, nonetheless, not stable in the atmosphere and when exposing the
electroplated coating to air, hydrogen sulphide and aluminium oxide will form according to:
Al S 3H O → 3H S g Al O ∆G 79.185 at 20 ) (28)
This was verified with an odour of hydrogen sulphide when drying the electrodeposited
specimen in air.
Even though H, O, Cl and S contaminants were found in the electroplated aluminium

coating, an attempt was made to produce the desired Al3Ni2 structure via heat treatment as
before. The micrographs in Fig. 54 and Fig. 55 combined with the EDS analyses in Table 11
indicate that only a small amount of aluminium actually diffuses into the structure and forms
the Al3Ni2 intermetallic structure (phase 2 in Fig. 55). The remaining deposited aluminium
coating expands intensively during the heat treatment and a dark layer (with higher Z-
contrast) containing large amount of oxygen and traces of S and Cl is formed. These dark
layers most likely act as a diffusion barrier between the aluminium and nickel. It is,
therefore, evident that the incorporated impurities lead to crack formations and expansion of
the structure when heat treated and little or no interdiffusion between the aluminium and
nickel takes place. Due to improper Al3Ni2 formation from ionic liquid electroplating, no
further attempts for producing large surface area nickel structure were done this screening.
7.5 Physical vapour deposition of aluminium onto electroplated

sulfamate nickel
7.5.1 Introduction
For the porous nickel electrodes prepared during this project the effort has mainly been
concentrated on coating of a commercially pure nickel. For lowering the production cost for
large scale production, plating nickel onto an iron or a stainless steel plate would be a better
alternative. In this screening, PVD aluminium is deposited onto an electroplated nickel
substrate.
A sulfamate nickel coating was electroplated onto a 50×50 mm 304 stainless steel substrate
in an electrolyte containing ca. 300 g/l Ni(SO3NH2), 40 g/l H3BO3 and 10 g/l NiCl2×6H2O.
The deposition current density was 4.5 A/dm2 and the plating duration was 2 hours at 45°C.
The pre-treatment of the steel substrate was as follows; cathodic degreaser for 2 min.,
pickling for 1 min., etching in 20% HCl and 5% H2O2 for 1 min. (to remove chromium
oxides from the surface), followed by cathodic degreaser for 2 min. and pickling for 1 min.
In order to assure good adhesion, a thin layer of woods nickel was electroplated onto the
steel substrate according to; anodic at 7 A/dm2 for 2 min (to remove the last oxides on the
surface) and cathodic deposition at 14 A/dm2 for 2 min.
The nickel plated steel substrates were PVD treated with about 20 µm aluminium as
explained in section 7.1.2. The aluminium PVD specimens were subsequently heat treated in
an atmospheric furnace for 4 hours at 610°C followed by selective aluminium leaching in 30
wt.% KOH and 10 wt.% KNaC4H4O6×4H2O at 80°C with stirring for 24 hours.
Sulfamate nickel coatings are known to have low stress and high ductility. These properties
could benefit the durability of the nickel electrodes and therefore sulfamate nickel coating
was selected over other nickel coatings. Fig. 56 and Fig. 57 show SEM micrographs of the
PVD aluminium nickel structure after 4 hours of heat treatment and after heat treatment
followed by selective aluminium leaching. Here shorter heat treatment is selected due to
knowledge gained from other thermo-chemical diffusion processes investigated; see
Appended papers II and III. The results from the EDS analyses are listed in Table 12.
72
Fig. 56: Cross section SEM micrographs of PVD aluminium on sulfamate nickel substrate heat treated at 610°C
for 4 hours.
Fig. 57: Cross section SEM micrographs of PVD aluminium on sulfamate Ni substrate heat treated at 610°C for 4
hours and alkaline aluminium leached.
Name of phase PVD Al on sulfamate Ni PVD Al on sulfamate Ni

/original phase heat treated heat treated and leached
Al Ni Al Ni O
[wt.%] [wt.%] [wt.%] [wt.%] [wt.%]
Al3N2 39 61 7 71 22
AlNi 29 71 32 68 -
AlNi3 16 84 16 84 -
Ni - 100 - 100 -
Table 12: EDS analyses on the PVD Al on sulfamate Ni specimens (Fig. 56and Fig. 57). All elements from the
periodic table except for carbon are analysed.
It is evident from the SEM micrographs and the EDS analysis that the whole Al3Ni2 phase is
leached during the leaching process. Porous nickel coating containing 7% aluminium is
developed by this process technique. The high oxygen content in the porous layer is most
likely due to resin trapped inside the pores, which supports the proposal that the structure is
porous. No cracks are visible in the porous structure.
7.6 Conclusions on the manufacturing of a high surface area nickel

coating
In the present work an attempt was made to produce dense high surface area nickel structure
using five different process techniques. The results together with the advantages,
disadvantages are summarised in Table 13. The results show that due to the high affinity of
aluminium towards hydrogen and oxygen, production of oxide free Al3Ni2 alloy coatings in
an inexpensive and simple manner is challenging. The next step in the development process
is AWE efficiency and durability testing of the electrodes produced by direct diffusion
between aluminium and nickel sheets and the PVD aluminium onto electroplated sulfamate
nickel.
Process Advantages Disadvantages Results

technique
PVD Al onto Ni Even and well PVD is a relatively expensive Producing high surface area nickel
levelled porous process technique. electrodes by PVD of Al onto a Ni
nickel coating is plate was found to be successive
formed. and numerous efficiency, durability
and metallography investigations
have been performed, see chapter 8,
Appended papers I,I and II.
Hot dip Al of Ni Inexpensive Hydrogen voids and oxide When heat treated about 100 µm of
plates and simple inclusions are observed. Al3Ni2 was formed. The
Dissolution of the nickel coupon intermetallic coating contained
during immersion. cracks and insulating oxides making
Inhomogeneous wetting of it unsuitable electrocatalytic coating
aluminium and difficulties to for AWE.
control the thickness of the
aluminium coating.
Thermo- Inexpensive and Operating in argon atmosphere is A dense macro- and supposedly
chemical simple. No post a complication factor for large nanoporous coating of about 100
diffusion of Al heat treatment is scale production. µm was formed. Crack formation
and Ni sheets needed. perpendicular to the surface increase
the risk of gas erosion corrosion
during AWE operation.
Al ionic liquid Inexpensive Uneven aluminium coating. The impurity inclusions
electroplating Contains impurity inclusions such incorporated in the as deposited
as chlorine, sulphur and oxides. structure makes it unsuitable as an
electrode coating for industrial
AWE
PVD Al onto PVD is a Less expensive than using solid Producing high surface area nickel
electroplated relatively nickel plate. electrodes by PVD of Al onto a
sulfamate Ni expensive sulfamate Ni was found to be
process successive.
technique
.
Table 13: Summary from screening of process techniques for producing high surface nickel electrocatalyst
74
8 Efficiency and Durability Measurements on

the Developed PVD Al/Ni Electrodes
8.1 Introduction
In order to characterise and evaluate the nickel supported PVD Al/Ni electrocatalyst,
numerous electrochemical and durability measurements have been performed. Here selected
tests will be introduced and discussed in order to evaluate the overall quality of the
electrodes. A part of these results have been published or are to be subjected to publication,
see Appended papers I and II
8.2 Experimental procedure
The electrodes to be tested were produced by the means of PVD of aluminium onto a nickel
substrate as reported in section 7.1. Cathodic measurements of electrodes produced in the
same manner, except for reduced heat treatment, are included for comparison. These
electrodes were heat treated for 10, 20 and 30 min. instead of the 24 hours. Structural,
morphology and composition investigations are given in Appended papers I and II.
The activity and durability of the electrodes was evaluated by means of potentiodynamic
measurements, single cell electrolysis measurements and measurements carried out in an
electrolysis stack. The potentiodynamic measurements were carried out using the half-cell
test setup introduced in section 5.3. The potential of the working electrode was recorded
against an Hg/HgO reference electrode from Radiometer (XR440) and a counter electrode
made of nickel. The electrolyte was 1 M KOH and the measurements were carried out at
25°C. The half-cell measurements were IR-compensated by means of current interruption.
All following half-cell potentials reported are determined versus the Hg/HgO reference
electrode.
The cathodic potentiodynamic measurements were carried out as follows: conditioning at

-0.8 or -0.9 V for 30 minutes in order to eliminate the amount of hydrates from the surface,
open circuit potential (OCP) until 0.01 mV/sec stability followed was reached, followed by a
cathodic sweep from -0.615to -1.415 V using scan rate of 1 mV/s. Potentiodynamic
measurements on polished nickel was performed in the same manner for comparison. In
order to capture potential outputs in the same range as for the developed electrodes the upper
limit voltage for the polished nickel measurements was increased to -1.715 V.
In the case of anodic measurements the procedure was initialised by a 30 min. conditioning
at 0.815 V, in order to facilitate a stable NiO structure, followed by OCP until 0.01 mV/sec
stability was reached. The potentiodynamic measurements were recorded from 0.4 to 0.9 V
at a scan rate of 1 mV/s.
Efficiency and Durability Measurements 75
Durability and whole cell efficiency investigations were carried out in the electrolysis test
setup introduced in section 5.2. The electrolyte contained 50 wt.% KOH and the
measurements were operated at 120C. The electrolyte was pre-electrolysed at 2 V potential
for 3 days prior to testing. The high concentration of the electrolyte was in order to achieve
high conductivity and to avoid the electrolyte from boiling. The diaphragm used was a
Zirfon® Perl 500 UTP. The distance between the anode and cathode was 2.5 mm, 1 mm
from the diaphragm on each side plus the thickness of the diaphragm which is approximately
0.5 mm.
Stack measurements were performed in a 17 cell bipolar, non-zero gap electrolysis stack
introduced in section 5.4. The stack contained 30 wt.% KOH and was operated under altered
conditions, at a maximum temperature of 80C and a pressure of 22 bar for approximately
9000 hours. Operating data was captured during the first month of operation and the
approximately 9000 hours of operation. The data management was carried out at
GreenHydrogen.dk
8.3 Results and discussions
8.3.1 Half-cell measurements
8.3.1.1 Cathodes
Results from the potentiodynamic measurements recorded on the developed PVD Al/Ni
electrodes produced with different duration of heat treatment are shown in Fig. 58.
Fig. 58: Cathodic potentiodynamic polarisation curves recorded on the developed electrocatalyst produced with
different heat treatments (10, 20, 30 minutes and 24 hours) compared to polished nickel. The electrolyte contains
1 M KOH and the experiments are performed at 25C.
76
For the practise of finding the exchange current density and calculating the Tafel slopes for
the reactions, the results are re-plotted in logarithmic scale (Fig. 59) where the x-axis
intersects the y-axis at the theoretical potential of HER (-0.943 V).
Fig. 59: Re-plot of the cathodic potentiodynamic curves from fig. Fig. 58 in order to find I0 and Tafel slopes. The
grey lines indicate Tafel slope 1 from 0.02 to 0.1 A/cm-2.
Results from the cathodic potentiodynamic measurements are gathered in Table 14.
Tafel slope 1 Tafel slope 2 I0 ηHER ηrev

[mV/dec.] [mV/dec.] [mA/ cm2] @ 200 A/cm2 @ 200 A/cm2*
[mV]
Polished Ni 141 171 4e-2 527 64%
PVD Al/Ni 102 245 3 184 84%
10min.
PVD Al/Ni 94 225 5 157 86%
20min.
PVD Al/Ni 83 228 6 142 87%
30min.
PVD Al/Ni 111 167 5 162 85%
24h.
Table 14: Tafel slopes, HER overpotential (ηHER) and calculated efficiency (ηref) from the cathodic
potentiodynamic measurements recorded on the developed electrode.
*Calculated according to the reversible potential of HER (-943 mV vs. Hg/HgO).
The Tafel slope is calculated from two current density intervals. Tafel slope 1 is calculated
from the range of 20-100 mA/cm2 and Tafel slope 2 from the interval of 100-1000 mA/cm2.
The results from the cathodic potentiodynamic measurements together with efficiency
calculations are collected in Table 14.
The HER overpotential at 200 mA/cm2 is up to 385 mV less for the developed electrocatalyst
compared to polished nickel. The electrode showing the best catalytic activity is the PVD
Al/Ni heat treated for 30 minutes, comprising a Tafel slope of 83 mV/dec. and exchange
current density for the hydrogen production of 6 mA/cm2. At higher current density the 24
hours heat treated electrode outpaces the others with a Tafel slope of 167 mV/dec. compared
to the 228 mV/dec. for the 30 minutes heat treated electrode. Accordingly, the 24 hours
electrode is more applicable to high current density operations. The reason could be that the
electrocatalytic structure formed with the 24 hours heat treatment is better to release the
hydrogen gas bubbles from the electrode surface. The efficiency of all electrodes range from
84 to 87% compared to the reversible potential of HER.
8.3.1.2 Anodes
The electrodes heat treated for 10, 20 and 30 minutes were not found to be stable as anodes
during potentiodynamic measurements, see Appended paper II. Accordingly, results from
these will not be analysed.
Two plots, with and without logarithmic scale, are prepared for the anodic potentiodynamic
measurements, see Fig. 60 and 61.
Fig. 60: Anodic potentiodynamic polarisation curves recorded on the developed electrocatalyst compared to
polished nickel.
78
Fig. 61: Re-plot of the anodic potentiodynamic curves from Fig. 60 in order to find I0 and Tafel slopes. The grey
lines indicate Tafel slope 1 from 0.02 to 0.1 A/cm-2.
For the logarithmic plot, the x-axis intersects the y-axis at the theoretical potential for OER
(286 mV). The Tafel slopes are calculated from the current density ranging from 0.02-0.1
A/cm2. Results from the anodic potentiodynamic measurements are listed in Table 15.
Tafel slope 1 I0 ηOER ηrev

[mV/ dec] [mA/cm2] @ 200 mA/cm2 @200 mA/cm2*
[mV]
Polished Ni 60 2e-5 437 39%
PVD Al/Ni 24h 63 3e-6 387 42%
Table 15: Tafel slopes, OER overpotential (ηOER) and calculated efficiency (ηref) from the anodic
potentiodynamic measurements recorded on the developed electrode.
* Calculated according to the theoretical potential of OER is estimated to be 286 mV vs. Hg/HgO.
The developed electrocatalyst is shown to have only 50 mV less OER overpotential

compared to polished nickel. Evidently, the developed electrocatalytic structure does not
compose superior catalytic behaviour towards OER.
8.3.2 Durability measurements in an electrolysis cell
The electrolysis cell measurements were carried out by applying the PVD Al/Ni electrodes,
prepared by 24 hours heat treatment, both as anode and cathode. First the electrolysis cell
was operated continuously for over 1500 hours. During the 1500 hours a couple of
shutdowns occurred. After each shut-down, the electrolysis measurements were started again
without making any changes to the electrolysis setup. Thereafter, the electrolyte was
changed out for a new fresh one. The data captured during the durability test compared to
durability measurements prepared on polished nickel electrodes are shown in Fig. 62.
Efficiency calculations after different time of operation are collected in Table 16.
Fig. 62: Electrolysis whole cell measurements, where the developed electrode is applied both as anode and
cathode, operated at 200 mA/cm2, 120C in 50%KOH. Electrolysis measurements operated under the same
conditions, where polished nickel is applied as anode and cathode, is plotted for comparison.
ηCell ηCell ηCell ηCell ηCell ηCell ηCell

@ @2 @ 200 @ 500 @ 1500 @ 1550 h. @ 1600 h.
5 min. h. h. h. h. new new
electrolyte* electrolyte
Polished Ni 79% 78% 75% 74% - -
PVD Al/Ni 91% 88% 84% 83% 83% 86% 84%
24h
Table 16: Efficiency calculations for electrolysis durability test, operated at 200 mA/cm2, 120C and 50 wt%
KOH, recorded on the developed electrodes compared to polished nickel. The efficiency values are calculated
according to the HHV. *Measured immediately after change of electrolyte.
During the first minutes of operation the potential between the electrodes is measured to be
about 1.6 V, resulting in efficiency values above 90%. Unfortunately, the potential increases
continually until about 100 operating hours, where the cell efficiency has dropped to 83-
84%. Thereafter the potential is stable during the rest of the 1500 operating hours. The most
obvious reason for the efficiency drop during operation is degradation of the electrodes by
gas-erosion. That implies that some of the porous structure is “blown off“ during operation
resulting in less electrocatalytic surface area available for the electrolysis reactions to take
place. This assumption can also, to some extend, be verified by the black powder found on
the bottom of the electrolysis cell after operation, see Fig. 63. It is noted that such sediments
were not observed for any of the potentiodynamic measurements presented above.
80
Fig. 63: Left: Black particles found in the electrolyte after over 1600 time of durability testing. Right: the
electrolyte in the electrolysis cell after short duration of testing with a new electrolyte.
However, due to the fact that the same degradation trend of deactivation is observed for the
polished nickel electrodes there must be an additional explanation of the degradation.
Formation of nickel hydrides inside the electrode lattice has been described by various
authors [115]–[118]. The formation of nickel hydrides changes the electronic configuration
of the electrocatalyst from d-character to sp-character by filling the d-band. This electronic
configuration is similar to copper and silver, which are known to have higher hydrogen
overpotential compared to nickel [118]. Hence, the 5% efficiency decrease during the 500
hours of operation for polished nickel can most probably be related to the formation of nickel
hydrides. Accordingly, some part of the efficiency loss observed for the durability testing of
the PVD Al/Ni electrode is supposed to originate from the formation of nickel hydride
diminishing the electrocatalytic activity of the catalyst towards the HER.
Another factor that could influence the increased potential is the evaporation of KOH with
water resulting in lower ionic conductivity of the electrolyte. Indeed, during operation white
KOH crystals are detected on the gas outlet pipes. The evaporation of water was
compensated during operation with addition of pure water, but no additional KOH is added
during the operating period. Hence, if the KOH evaporates with the water, the ionic
conductivity in the electrolyte will obviously decrease leading to higher ohmic drop between
the anode and cathode. In order to investigate this, the old electrolyte was changed out for a
new fresh 50 wt.% KOH after approximately 1550 hours of operation. Immediately after the
changing of the electrolyte the efficiency value increased from 84% to 86%. After less than
10 hours the efficiency reached its initial value of 84%. These results are understood in that
way that the evaporation of KOH is not in that large quantity that the factor diminishes the
ionic conductivity between the electrodes.
Regarding the ohmic drop between the two electrodes it should be mentioned that no
external electrolyte flow is applied to the single cell setup, as usually done in commercial
electrolysis stacks to enhance bubble separation from the electrode surfaces during
operation. This means that gas bubbles cover the electrode surfaces at all times resulting in
less available electrode surface area and higher ohmic drop between the electrodes.
Commercial electrolysers are also often pressurised in order to minimise the volume of the
gas bubbles in the electrolyte and, thus, lowering the ohmic drop caused by them. It is
therefore expected that the ohmic drop between the electrodes in the single cell electrolysis
setup is somewhat larger than for industrial electrolysers.
During the durability testing of the electrodes several shutdowns occurred. The reasons for
the shutdowns were of technical kind and will not be emphasised here. What is noteworthy is
that no further deactivation of the PVD Al/Ni electrocatalyst was detected after the
shutdowns. The resistivity to shutdowns is an important factor in the development of
electrocatalysts.
Fig. 64 shows the PVD Al/Ni electrodes after the durability testing. Although evidently some
of the electrocatalytic porous nickel structure has scaled off during the procedure, the
electrodes still appear black. The anode clearly seems more degraded than the cathode.
Fig. 64: The PVD Al/Ni electrodes used for single cell electrolysis durability testing after over 1600 hours of
operation. Left: Cathode. Right: Anode. The electrode surfaces still appear black indicating that some remaining
high surface area skeletal nickel coating.
In order to investigate the influence of the durability testing on the hydrogen electrode,
cathodic potentiodynamic polarisation measurements were performed. The result from the
potentiodynamic testing on the electrode applied as cathode in the single cell durability test
is shown in Fig. 65 and 66. The process parameters for the potentiodynamic measurements
are the same as described in the experimental chapter. The Tafel slope is calculated from two
current density intervals as before, Tafel slope 1 in the range of 20-100 mA/cm2 and Tafel
slope 2 in the range of 100-1000 mA/cm2. The results from the cathodic potentiodynamic
measurements together with efficiency calculations are gathered in Table 17.
82
Fig. 65: Cathodic potentiodynamic polarisation curves recorded on the PVD Al/Ni electrode applied as cathode
for over 1600 hours durability testing. The electrolyte contains 1 M KOH and the experiments are performed at
25C.
Fig. 66: Re-plot of the cathodic potentiodynamic curves in Fig. 65 in order to find I0 and the Tafel slope. The
dotted line indicate Tafel slope 1 from 0.01 to 0.1 A/cm-2.
Tafel slope 1 Tafel slope 2 I0 ΗHER ηrev

[mV/dec] [mV/dec] [mA/ cm2] @ 200 @200 mA/cm2*
mA/cm2[mV]
PVD Al/Ni 24h 140 300 7 218 81%
after 1600 h. of
operation
Table 17: Results and calculated efficiency (η) from the cathodic potentiodynamic measurements recorded on
electrode applied as cathode for over 1600 hours single cell electrolysis testing. *Calculated according to the
theoretical potential of HER ( -943 mV vs. Hg/HgO).
The post potentiodynamic measurement reveal that the PVD Al/Ni has 56 mV higher
overpotential towards HER at 200 mA/cm2 current density compared to a fresh PVD Al/Ni
electrode. This results in about 4% reduction in efficiency. An increase of 29 mV/dec. and
133 mV/dec. is observed for Tafel slope 1 and Tafel slope 2, respectively. The great
difference between Tafel slope 2 for the fresh and the durability tested electrodes indicate
that a reduced surface area has a more significant effect at higher current densities compared
to lower current densities. The increased overpotential for the HER originates form a
combination of less actual surface area and nickel hydrides that have been formed in the
lattice of the electrocatalyst.
8.3.3 Electrolysis stack measurements
16 large electrodes with the developed PVD Al/Ni 24 h. coatings on both sides were
produced for testing in a 17 cell bipolar alkaline electrolysis stack. Together with serving as
a durability test stack for the PVD Al/Ni electrodes the stack was used for the development
and demonstration project called H2-College in Herning. Hence, the electrolysis stack was
located on site where the hydrogen was to be produced and not in the laboratory. Continuous
potential vs. current measurements were therefore not made on the stack during operation.
Instead measurements where prepared occasionally during the operation period.
The alkaline electrolysis stack was at that time in the development stage and for safety
reasons, the stack was operated under low temperature and pressure in the beginning of the
test period. As it often is for projects in the development stage the biggest focus was on
making all the coupled mechanisms to work and function together. Consequently, rather few
operation parameters were captured at that time. The limited amount of data recorded during
the first month of operation is shown in Fig. 67.
84
Fig. 67: Current density vs. cell voltage recorded on the 17 cell electrolysis stack during the first month of
operation. The operation conditions were 40-45C and 10 bar.
The electrolysis stack was operated under altering conditions for over 9000 hours. Thereafter
plentiful of efficiency measurements were performed on the stack. Both the dependence of
the voltage towards the current and the temperature were measured. Fig. 68 and 69 show the
current vs. voltage and temp. vs. voltage data captured after 9000 hours of operation. For
making the comparison easier between the previous measurements on the PVD Al/Ni
electrodes and the stack measurements the measured stack voltage is divided by 17 in order
to achieve the average cell voltage.
Fig. 68: Current vs. voltage and efficiency data captured on the 17-cell bipolar electrolysis stack after
approximately 9000 operating hours. The stack was operated at 60C and 22 bar. The efficiency calculations are
based on the HHV.
Fig. 69: Temperature vs. voltage plot captured on the 17-cell bipolar electrolysis stack at 22 bar and 200
mA/cm2current density, after approximately 9000 operating hours, showing the temperature dependence of the
cell voltage. The efficiency calculations are based on the HHV.
A cell efficiency of about 81% is observed at 200 mA/cm2 at 80C. This is 3% points less
than the efficiency value observed after over 1600 hours of single cell durability test at
120C and 50 wt.% KOH. The efficiency values from the stack measurements are, however,
most likely lower than the actual cell efficiencies due to losses in the stack itself.
86
The test performed after 9000 hours of operation were focused on measuring the highest
operating values. Therefore, the data captured in the start and after the full operating time are
not fully comparable. In Table 18 three values from the stack measurements have been
chosen for durability assessment, one from the beginning of the test period and two from
after 9.000 hours. No serious deactivation during the operation period can be detected from
these values.
Temp. Pressure Stack Current Stack Cell Stack

[°C] [Bar] Current density Voltage Voltage efficiency
[A] [mA/cm2] [V] [V] [%]
First month of 40 10 40 150 33 1.98 75
the operation
period
After 9.000 h. 40 22 54 200 35 2.05 73
operation
After 9.000 h. 60 22 40 150 32 1.88 79
operation
Table 18: Selected data from Fig. 67, Fig. 68 and Fig. 69 for durability assessment. The efficiency calculations
are based on the HHV.
8.4 Conclusions for efficiency and durability testing
The developed electrocatalyst is found to have up to 385 mV less hydrogen overpotential

and 50 mV less oxygen overpotential, compared to polished nickel measured at 200 mA/cm2
and 25C. Longer heat treatment, 24 hours compared to 30 min., results in more active
hydrogen catalyst at higher current densities, resulting in a Tafel slope of 167 mV/dec. for
the 24 hours heat treated electrocatalyst compare to 228 mV/dec. for the 30 min. heat treated
electrocatalyst
According to the half- cell measurements, the efficiency of an ideal electrolysis cell, where
the resistance between the electrodes can be neglected, operated with the PVD Al/Ni 24 h.
electrocatalysts, both as anode and cathode, is found to be 83% [HHV], measured at 200
mA/cm2, 25C in 1M KOH.
Single cell electrolysis measurements, carried out in 50 wt.% KOH, at 120C and 200
mA/cm2, present above 88% cell efficiency during the first two hours of operation. The cell
efficiency decreases to 84% after about 100 operating hours and is thereafter constant
throughout the remaining 1500 hours of operation. The reason for the drop in efficiency
during time is proposed to degradation of the electrodes together with formation of nickel
hydrides in the nickel lattice. Operating at such high temperature and strong alkaline
electrolyte some efficiency loss must be expected.
In a pressurised electrolysis stack with external convection system, the boiling point of the
electrolyte increases, due to the pressure, and less KOH can be applied. Moreover, the
volume of the gas bubbles formed will decrease under the pressure and the ohmic loss
between the electrodes will become less due to the convection. Consequently, higher
efficiency and less degradation might be expected for the developed structure operated in an
optimised electrolysis stack.
Durability testing of the developed PVD Al/Ni 24 h. electrocatalyst carried out in a 17 cell
bipolar electrolysis stack containing 30 wt.% KOH, operated under altered conditions, at a
maximum temperature of 80C and a pressure of 22 bar, for approximately 9000 hours,
indicates no deactivation of the electrodes during the operation period. A stack efficiency of
81% is measured at 200 mA/cm2 at 80C. The electrolysis test stack is in the development
stage and some losses could be expected in the stack itself.
88
9 Conclusions
In a world of ever increasing energy demand [4], shortage of fossil fuels [151] and stronger
legalisations towards reduction of CO2 emission [152], new alternative fuels for domestic
heat and power as well as for transportation are required. Renewable energy is commonly
recognised to be the answer for a more secure, reliable and sustainable future. However, in
order to keep production and consumption in balance, more load management is needed.
Production of hydrogen by water electrolysis can serve as a balancing load and at the same
time act as energy storage.
The water electrolysis technique is not a new invention and, in fact, the technique was
developed over two centuries ago [44]. Currently there are two types of commercially
available water electrolysis systems. These are: alkaline water electrolysis (AWE) and
polymer electrolyte membrane (PEM) electrolysis. The available production rate for AWE is
up to 25 times larger than for PEM. Still, PEM electrolysis systems are at least 10 times
more expensive than AWE and can offer only half of the life time of AWE. Consequently,
alkaline electrolysis is the current standard for large-scale hydrogen production among water
decomposition techniques.
Commercial alkaline electrolysers operates in a liquid electrolyte containing 25-30% KOH at

a temperature ranging from 80 - 90°C and current density in the range of 100-400 mA/dm2
[10], [16],[49]. Efficiency values up to 82% have been reported [46]. One of the largest cell
efficiency losses for alkaline electrolysis originates from the activation energies for hydrogen
and oxygen, generated at the electrodes. The activity of an electrocatalyst depends on the
electron configuration of the catalyst material, the structure and geometry and the actual
surface area. Among non-noble metals, nickel is one of the most stable in strong alkaline
solutions. Nickel is also a relatively good catalyst for hydrogen and oxygen formation.
Nickel or nickel plated substrates are therefore typically the core material used in electrodes
for AWE systems [6],[10][45].
A great deal of work has been devoted to the development of electrodes for AWE during the
past 90 years. The state-of-the-art electrodes have, however, not changed much during the
years. Among the newly developed electrocatalysts, durability measurements are usually
lacking and few of the published electrocatalysts have actually been tested at current
densities applicable for industrial AWE. Hence, finding low cost electrode materials that are
both efficient and having long term stability is one of the remaining challenges within the
field of AWE.
In the present PhD study the core aim has been to develop high surface area nickel electrodes
that are both efficient and durable in large scale alkaline water electrolysis systems. For
industrial applications bipolar electrolysers are more commonly applied than monopolar. In
bipolar configuration each electrode serves as an anode, on the one side and a cathode, on
the other. Hence, having the same electrocatalytic surface on both sides of the electrodes
Conclusions 89
increases the simplicity and decrease the production cost of the electrodes. Although the
main focus was kept on the developing electrodes with high electrocatalytic activity towards
the HER, the efficiency of the developed electrodes towards the OER was also investigated.
For reliable electrochemical and durability investigations new measurement setup have been
developed and constructed during this PhD study.
In the present work, new hydrogen electrodes were developed. The electrodes were produced
by physical vapour depositing (PVD) of about 20 µm aluminium coating onto a nickel
substrate. The PVD Al-Ni couples were heat treated for 24 hours in order to facilitate large
crystalline Al3Ni2 alloy formation. Thereafter the electrode coating, consisting primarily of
the Al3Ni2 phase, was selective aluminium alkaline leached. The developed electrocatalytic
surface was characterised with a large actual surface area, uniform structure and with good
interlayer adhesion, which is critical for industrial application. IR-compensated polarisation
curves prepared in a half-cell setup with 1 M KOH electrolyte at 200 mA/cm2 at room
temperature reveals that 435 mV less potential is needed to decompose water into hydrogen
and oxygen with the developed porous PVD Al/Ni electrodes as compared to solid nickel
electrodes. The hydrogen electrodes were measured to have 85% efficiency under the same
conditions.
By increasing the operating temperature in AWE systems the kinetics of the hydrogen
production can be increased. Non-zero gap, single cell electrolysis measurements were
carried out where the developed PVD Al/Ni electrodes were applied both as cathode and
anode. The operation conditions were; 50 wt.% KOH, 120C and 200 mA/cm2. The cell
efficiency was measured to be 88% (HHV) during the first two hours of operation. The cell
efficiency decreased to 84% after about 100 operating hours and was thereafter constant
throughout the remaining 1500 hours of operation. The reason for the efficiency drop over
time is proposed to be degradation of the electrodes together with formation of nickel
hydrides in the nickel lattice. The material degradation was found to be considerably more
severe at the anode compared to the cathode. However, operating at such high temperature
and strong alkaline electrolyte some efficiency loss must be expected.
16 full size electrodes with the developed PVD Al/Ni structure were produced for bipolar,
non-zero gap, industrial AWE stack. The electrodes were prepared with the developed high
surface area Ni structure on both sides. The electrolysis stack was operated with surplus
power from wind turbines and the hydrogen produced used to power 66 houses at the
campus of Århus University in Herning. The operation conditions of the electrolyser were;
30 wt.% KOH, maximum temperature of 80C and a pressure of 22 bar. The electrodes were
operated in the electrolysis stack for approximately 9000 hours. Comparing operating data
captured in first month of operation to data captured after over 9000 hours, indicates no
deactivation of the electrodes during the operation period. The stack efficiency after over
9000 hours of operation at 200 mA/cm2 at 80 C was measured to be 81% (HHV). It is noted
that the electrolysis test stack was in the development stage and some losses were expected
in the stack itself.
Microstructure investigations on the PVD Al-Ni diffusion couples at 610C, for various
times of heat treatments, indicate that the initial diffusion mechanism is dominated by grain
90
boundary diffusion of Ni-rich phases into the PVD Al structure. It is proposed that the first
intermetallic phase to form is AlNi3, appearing as small particles in the grain boundaries of
the columnar aluminium structure. Due to the high mobility of aluminium at the annealed
temperature, finding Ni and Ni-rich containing species to be the most mobile during the heat
treatment is highly unexpected and in contrast with other findings in the literature. The
diffusion mechanism can be the key to the good properties of the developed PVD Al/Ni
electrode.
Both the interdiffusion and the leaching procedures facilitate internal stresses in the
electrocatalytic coatings prepared in the present manner. Leaching of PVD Al-Ni structure
after a short time of diffusion, 10-30 minutes, results in formation of cracks perpendicular to
the Ni substrate. The cracks diminish the mechanical strength of the coating. PVD Al/Ni
electrodes heat treated for short times are found not to be stable under OER. Longer heat
treatments, up to 24 hours, result in grain growth of the leachable Al3Ni2 phase. Selectively
aluminium leaching of electrodes heat treated for 24 hours results in dense, crack free and
more mechanical stable structure. It is noteworthy that the electrodes heat treated for shorter
times are more prone to alkaline leaching. This is verified by the EDS analyses where only
4-5 wt.% aluminium residue is found in the leached structure of the specimens heat treated
for 10,20 and 30 minutes, compared to up to 15 wt.% aluminium found in the leached
electrode heat treaded for 24 hours. Also from the XRD analyses, some remaining Al3Ni2
peaks are observed from the leached structure heat treated for 24 hour where only pure
nickel peaks are observed from the leached 10-30 minutes heat treated structures.
Because of the partial penetration of the current into the deeper pores of the structure one
could expect that only a limited fraction of the actual surface area, of the highly dispersed Ni
structure, contributes to the electrochemical reaction during electrolysis. Potentiodynamic
measurements prepared on PVD Al/Ni electrode surfaces heat treated for 10, 20 and 30
minutes, resulting in different thicknesses of porous Ni surface, indicate that the
electrocatalytic activity increases in proportion to the porous layer thickens up to the whole
20 µm tested. However, in order to optimise the electrochemical and mechanical properties
of the developed electrocatalytic surface, more thorough investigations on the effect of
leaching and heat treatment parameters are essential.
After the successful experience of the first generation of electrodes for AWE, four other
process techniques, and combinations of these, for producing similar electrodes were
assessed. The process techniques were selected with the aim of reducing the production cost
associated with the electrode manufacture. The screened processes were; hot dip aluminising
of nickel followed by thermo-chemical diffusion, direct thermo-chemical diffusion of
aluminium and nickel sheets, aluminium ionic liquid electroplating on a nickel plate
followed by thermo-chemical diffusion and physical vapour deposition of aluminium onto
electroplated sulfamate nickel substrate followed by thermo-chemical diffusion. Due the
high affinity of aluminium towards hydrogen and oxygen, producing oxide free Al3Ni2 alloy
coatings in an inexpensive and simple manner was found to be challenging. Only the direct
diffusion between aluminium and nickel sheets in argon atmosphere and the PVD Al onto
electroplated sulfamate nickel were found to give promising coatings. These surfaces have,
yet, not been tested electrochemically.
Bibliography 91
Bibliography
[1] WMO. Greenhouse Gas Bulletin. The State of Greenhouse Gases in the Atmosphere.
Based on Global Observations through 2012, [Online] November 2013,
http://www.wmo.int/pages/prog/arep/gaw/ghg/documents/GHG_Bulletin_No.9_en.pd
f (Accessed: 12 November 2013).
[2] United Nations. Framework Convention on Climate Change. Report of the
Conference of the Parties on its sixteenth session, held in Cancun from 29 November
to 10 December 2010, [Online] 15 March 2011,
http://unfccc.int/resource/docs/2010/cop16/eng/07a01.pdf#page=2 (Accessed: 12
November 2013).
[3] European Commission. Energy 2020. A strategy for competitive, sustainable and
secure energy, [Online] November 2012, http://eur-
lex.europa.eu/LexUriServ/LexUriServ.do?uri=COM:2010:0639:FIN:EN:PDF
(Accessed: 12 November 2013).
[4] OECD, World Energy Outlook 2013. Paris: Organisation for Economic Co-operation
and Development, 2013
[5] Fuel Cells Today, Water Electrolysis & Renewable Energy, [Online] May 2013,
http://www.fuelcelltoday.com/media/1871508/water_electrolysis___renewable_energ
y_systems.pdf (Accessed: 12 November 2013).
[6] T. Smolinka, ‘Fuels – Hydrogen Production | Water Electrolysis’, in Encyclopedia of
Electrochemical Power Sources, Jürgen Garche, Ed. Amsterdam: Elsevier, 2009, pp.
394–413.
[7] D. A. J. Rand and R. M. Dell, Hydrogen Energy : Challenges and Prospects RSC
Energy Series v. 1. Royal Society Of Chemistry.
[8] S. S. Zumdahl, in Chemical principles, 4th ed., 2002, pp. 470–471.
[9] H. Wendt, in Electrochemical hydrogen technologies: electrochemical production
and combustion of hydrogen, Elsevier, 1990, pp. 139–142.
[10] K. Zeng and D. Zhang, ‘Recent progress in alkaline water electrolysis for hydrogen
production and applications’, Progress in Energy and Combustion Science, vol. 36,
no. 3, pp. 307–326, Jun. 2010.
[11] S. Licht, ‘Thermochemical and Thermal/Photo Hybrid Solar Water Splitting’, in
Solar Hydrogen Generation, K. Rajeshwar, R. McConnell, and S. Licht, Eds.
Springer New York, 2008, pp. 87–121.
[12] I. Abe, ‘Hydrogen Production from Water’, in Energy Carriers and Conversion
Systems, EOLSS Publishers Company Limited, 2009.
[13] E. Bardal, in Corrosion and Protection, Springer, 2004, pp. 35–44.
[14] M. Wang, Z. Wang, X. Gong, and Z. Guo, ‘The intensification technologies to water
electrolysis for hydrogen production – A review’, Renewable and Sustainable Energy
Reviews, vol. 29, pp. 573–588, Jan. 2014.
[15] H. Wendt and G. Imarisio, ‘Nine years of research and development on advanced
water electrolysis. A review of the research programme of the Commission of the
European Communities’, J Appl Electrochem, vol. 18, no. 1, pp. 1–14, Jan. 1988.
[16] K. Harrison and J. I. Levene, ‘Electrolysis of Water’, in Solar Hydrogen Generation,
K. Rajeshwar, R. McConnell, and S. Licht, Eds. Springer New York, 2008, pp. 41–
63.
[17] D. Pletcher, in Instrumental Methods in Electrochemistry, Ellis Horwood, 2001, pp.
24–25.
92
[18] S. Trasatti, ‘Electrochemical Theory | Hydrogen Evolution’, in Encyclopedia of

Electrochemical Power Sources, J. Garche, Ed. Amsterdam: Elsevier, 2009, pp. 41–
48.
[19] D. Pletcher, Instrumental Methods in Electrochemistry. Ellis Horwood, 2001.
[20] S. G. Neophytides, S. Zafeiratos, G. D. Papakonstantinou, J. M. Jaksic, F. E.
Paloukis, and M. M. Jaksic, ‘Extended Brewer hypo–hyper--interionic bonding theory
— I. Theoretical considerations and examples for its experimental confirmation’,
International Journal of Hydrogen Energy, vol. 30, no. 2, pp. 131–147, Feb. 2005.
[21] M. M. Jakšić, ‘Electrocatalysis of hydrogen evolution in the light of the brewer—
engel theory for bonding in metals and intermetallic phases’, Electrochimica Acta,
vol. 29, no. 11, pp. 1539–1550, Nov. 1984.
[22] M. M. Jaksic, ‘Hypo–hyper-d-electronic interactive nature of interionic synergism in
catalysis and
for hydrogen reactions’, International Journal of Hydrogen Energy, vol. 26, no. 6, pp. 559–
578, Jun. 2001.
[23] J. M. Jaksic, N. V. Krstajic, B. N. Grgur, and M. M. Jaksic, ‘Hydridic and
electrocatalytic properties of hypo-hyper-d-electronic combinations of transition
metal intermetallic phases’, International Journal of Hydrogen Energy, vol. 23, no. 8,
pp. 667–681, Aug. 1998.
[24] M. M. Jaksic, ‘Volcano plots along intermetallic hypo-hyper-d-electronic phase
diagrams and electrocatalysis for hydrogen electrode reactions’, Journal of New
Materials for Electrochemical Systems, vol. 3, no. 2, pp. 169–182, 2000.
[25] J. M. Jaksic, M. V. Vojnovic, and N. V. Krstajic, ‘Kinetic analysis of hydrogen
evolution at Ni-Mo alloy electrodes’, Electrochimica Acta, vol. 45, no. 25–26, pp.
4151–4158, Aug. 2000.
[26] G. S. Tasić, U. Lačnjevac, M. M. Tasić, M. M. Kaninski, V. M. Nikolić, D. L. Žugić,
and V. D. Jović, ‘Influence of electrodeposition parameters of Ni–W on Ni cathode
for alkaline water electrolyser’, International Journal of Hydrogen Energy, vol. 38,
no. 11, pp. 4291–4297, Apr. 2013.
[27] J. R. McKone, B. F. Sadtler, C. A. Werlang, N. S. Lewis, and H. B. Gray, ‘Ni–Mo
Nanopowders for Efficient Electrochemical Hydrogen Evolution’, ACS Catal., vol. 3,
no. 2, pp. 166–169, Feb. 2013.
[28] L. Birry and A. Lasia, ‘Studies of the hydrogen evolution reaction on Raney nickel-
molybdenum electrodes’, J. Appl. Electrochem., vol. 34, no. 7, pp. 735–749, Jul.
2004.
[29] J. Greeley, T. F. Jaramillo, J. Bonde, I. Chorkendorff, and J. K. Norskov,
‘Computational high-throughput screening of electrocatalytic materials for hydrogen
evolution’, Nat Mater, vol. 5, no. 11, pp. 909–913, Nov. 2006.
[30] S. M. A. Sillitto, N. J. E. Adkins, R. M. Ormerod, E. Paul, and D. R. Hodgson,
Characterisation of advanced Raney nickel electrocatalytic coatings produced by
the direct spraying method. Rugby: Inst Chemical Engineers, 1999.
[31] S. M. A. Sillitto, N. J. E. Adkins, D. R. Hodgson, E. Paul, and R. M. Ormerod,
‘Electrochemical testing and structural characterisation of nickel based catalytic
coatings produced by direct spraying’, in Advanced Catalytic Materials-1998, vol.
549, P. W. Lednor, D. A. Nagaki, and L. T. Thompson, Eds. Warrendale: Materials
Research Society, 1999, pp. 23–29.
[32] A. Kellenberger, N. Vaszilcsin, W. Brandl, and N. Duteanu, ‘Kinetics of hydrogen
evolution reaction on skeleton nickel and nickel-titanium electrodes obtained by
thermal arc spraying technique’, Int. J. Hydrog. Energy, vol. 32, no. 15, pp. 3258–
3265, Oct. 2007.
[33] A. Kellenberger, N. Vaszilcsin, and W. Brandl, ‘Roughness factor evaluation of
thermal arc sprayed skeleton nickel electrodes’, J. Solid State Electrochem., vol. 11,
no. 1, pp. 84–89, Nov. 2007.
Bibliography 93
[34] A. Kellenberger and N. Vaszilcsin, ‘The determination of the roughness factor of

skeleton nickel electrodes by cyclic voltammetry’, Rev. Chim., vol. 56, no. 7, pp.
712–715, Jul. 2005.
[35] G. Schiller, R. Henne, and V. Borck, ‘Vacuum plasma spraying of high-performance
electrodes for alkaline water electrolysis’, JTST, vol. 4, no. 2, pp. 185–194, 1995.
[36] J. Fournier, D. Miousse, and J. G. Legoux, ‘Wire-arc sprayed nickel based coating for
hydrogen evolution reaction in alkaline solutions’, Int. J. Hydrog. Energy, vol. 24,
no. 6, pp. 519–528, Jun. 1999.
[37] T. Boruciński, S. Rausch, and H. Wendt, ‘Raney nickel activated H2-cathodes Part II:
Correlation of morphology and effective catalytic activity of Raney-nickel coated
cathodes’, J Appl Electrochem, vol. 22, no. 11, pp. 1031–1038, Nov. 1992.
[38] V. S. Bagotsky, in Fuel Cells: Problems and Solutions, John Wiley & Sons, 2012, pp.
209–210.
[39] D. Pletcher, in Instrumental Methods in Electrochemistry, Ellis Horwood, 2001, pp.
168–171.
[40] P. Møller and L. P. Nielsen, in Advanced Surface Technology, 4th ed., Møller &
Nielsen, pp. 54–55.
[41] J. C. M. Keith B. Oldham, Electrochemical Science and Technology: Fundamentals
and Applications. John Wiley & Sons Inc.
[42] P. Møller and L. P. Nielsen, Advanced Surface Technology, 4th ed. Møller & Nielsen.
[43] A. J. Bard, Electrochemical methods : fundamentals and applications. 2001.
[44] P. Kurzweil, ‘History | Electrochemistry’, in Encyclopedia of Electrochemical Power
Sources, Jürgen Garche, Ed. Amsterdam: Elsevier, 2009, pp. 533–554.
[45] G. A. Edited By Luis M. Gandia, Renewable Hydrogen Technologies : Production,
Purification, Storage, Applications and Safety. Elsevier.
[46] A. Ursua, L. M. Gandia, and P. Sanchis, ‘Hydrogen Production From Water
Electrolysis: Current Status and Future Trends’, Proceedings of the IEEE, vol. 100,
no. 2, pp. 410 –426, Feb. 2012.
[47] R. L. LeRoy, ‘Industrial water electrolysis: Present and future’, International Journal
of Hydrogen Energy, vol. 8, no. 6, pp. 401–417, 1983.
[48] M. Carmo, D. L. Fritz, J. Mergel, and D. Stolten, ‘A comprehensive review on PEM
water electrolysis’, International Journal of Hydrogen Energy, vol. 38, no. 12, pp.
4901–4934, Apr. 2013.
[49] C. K. Dyer, P. T. Moseley, Z. Ogumi, D. A. J. Rand, and B. Scrosati, Encyclopedia of
Electrochemical Power Sources. Newnes, 2013.
[50] K. Scott, ‘Ion-exchange and charge driven processes’, in Industrial Membrane
Separation Technology, P. K. Scott and P. R. Hughes, Eds. Springer Netherlands,
1996, pp. 181–257.
[51] P. Vermeiren, W. Adriansens, J. P. Moreels, and R. Leysen, ‘Evaluation of the
Zirfon® separator for use in alkaline water electrolysis and Ni-H2 batteries’,
International Journal of Hydrogen Energy, vol. 23, no. 5, pp. 321–324, May 1998.
[52] H. Wendt and H. Hofmann, ‘Ceramic diaphragms for advanced alkaline water
electrolysis’, J Appl Electrochem, vol. 19, no. 4, pp. 605–610, Jul. 1989.
[53] R. B. Rebak, ‘Nickel Alloys for Corrosive Environments’, Advanced materials and
processes, vol. 157, no. PART 2, pp. 37–42, 2000.
[54] ‘Corrosion resistance of nickel and nickel-containing alloys in caustic soda and other
alkalies.’, INCO (Int Nickel Co, Inc) Corros Man, 1975.
[55] O. Pfleiderer, ‘Electrode’, 181857911-Aug-1931.
[56] K. A. Andreassen, H. Harang, J. B. Holte, and K. Widding, ‘Method for preparing
active cathodes for electrochemical processes’, US4171247 A16-Oct-1979.
[57] H. Vandenborre, P. Vermeiren, and R. Leysen, ‘Hydrogen evolution at nickel
sulphide cathodes in alkaline medium’, Electrochimica Acta, vol. 29, no. 3, pp. 297–
301, Mar. 1984.
94
[58] G. Hoogers, in Fuel Cell Technology Handbook, United States of America: CRC
Press, 2002, pp. 20–25.
[59] F. . Crnkovic, S. A. . Machado, and L. . Avaca, ‘Electrochemical and morphological
studies of electrodeposited Ni–Fe–Mo–Zn alloys tailored for water electrolysis’,
International Journal of Hydrogen Energy, vol. 29, no. 3, pp. 249–254, Mar. 2004.
[60] G. Sheela, ‘Zinc-nickel alloy electrodeposits for water electrolysis’, International
Journal of Hydrogen Energy, vol. 27, no. 6, pp. 627–633, 2002.
[61] W. K. Hu, ‘Electrocatalytic properties of new electrocatalysts for hydrogen evolution
in alkaline water electrolysis’, International Journal of Hydrogen Energy, vol. 25, no.
2, pp. 111–118, 2000.
[62] P. Los, ‘Hydrogen evolution reaction on Ni-Al electrodes’, Journal of Applied
Electrochemistry, vol. 23, no. 2, pp. 135–140, 1993.
[63] I. A. Raj, ‘Nickel-based, binary-composite electrocatalysts for the cathodes in the
energy-efficient industrial production of hydrogen from alkaline-water electrolytic
cells’, J Mater Sci, vol. 28, no. 16, pp. 4375–4382, Aug. 1993.
[64] Q. Han, S. Cui, N. Pu, J. Chen, K. Liu, and X. Wei, ‘A study on pulse plating
amorphous Ni–Mo alloy coating used as HER cathode in alkaline medium’,
International Journal of Hydrogen Energy, vol. 35, no. 11, pp. 5194–5201, Jun. 2010.
[65] N. Krstajic, V. Jovic, L. Gajickrstajic, B. Jovic, A. Antozzi, and G. Martelli,
‘Electrodeposition of Ni–Mo alloy coatings and their characterization as cathodes for
hydrogen evolution in sodium hydroxide solution’, International Journal of
Hydrogen Energy, vol. 33, no. 14, pp. 3676–3687, Jul. 2008.
[66] M. P. Marceta Kaninski, D. P. Saponjic, V. M. Nikolic, D. L. Zugic, and G. S. Tasic,
‘Energy consumption and stability of the Ni–Mo electrodes for the alkaline hydrogen
production at industrial conditions’, International Journal of Hydrogen Energy, vol.
36, no. 15, pp. 8864–8868, Jul. 2011.
[67] W. Hu, X. Cao, F. Wang, and Y. Zhang, ‘A novel cathode for alkaline water
electrolysis’, International Journal of Hydrogen Energy, vol. 22, no. 6, pp. 621–623,
Jun. 1997.
[68] R. N. Singh, L. Bahadur, J. P. Pandey, S. P. Singh, P. Chartier, and G. Poillerat,
‘Preparation and characterization of thin films of LaNiO3 for anode application in
alkaline water electrolysis’, J Appl Electrochem, vol. 24, no. 2, Feb. 1994.
[69] S. P. Singh, R. N. Singh, G. Poillearat, and P. Chartier, ‘Physicochemical and
electrochemical characterization of active films of LaNiO3 for use as anode in
alkaline water electrolysis’, International Journal of Hydrogen Energy, vol. 20, no. 3,
pp. 203–210, Mar. 1995.
[70] S. K. Tiwari, J. F. Koenig, G. Poillerat, P. Chartier, and R. N. Singh, ‘Electrocatalysis
of oxygen evolution/reductionon LaNiO3 prepared by a novel malic acid-aided
method’, Journal of Applied Electrochemistry, vol. 28, no. 1, pp. 114–119, Jan. 1998.
[71] J. O. Bockris and T. Otagawa, ‘The Electrocatalysis of Oxygen Evolution on
Perovskites’, J. Electrochem. Soc., vol. 131, no. 2, pp. 290–302, Feb. 1984.
[72] H. J. Miao and D. L. Piron, ‘Electrodeposition of Ni-transition alloys for the oxygen
evolution reaction’, J Appl Electrochem, vol. 21, no. 1, pp. 55–59, Jan. 1991.
[73] J. Balej, ‘Electrocatalysts for oxygen evolution in advanced water electrolysis’,
International Journal of Hydrogen Energy, vol. 10, no. 2, pp. 89–99, 1985.
[74] H. Wendt, H. Hofmann, and V. Plzak, ‘Materials research and development of
electrocatalysts for alkaline water electrolysis’, Materials Chemistry and Physics, vol.
22, no. 1–2, pp. 27–49, 1989.
[75] S. K. Tiwari, P. Chartier, and R. N. Singh, ‘Preparation of Perovskite‐Type Oxides of
Cobalt by the Malic Acid Aided Process and Their Electrocatalytic Surface Properties
in Relation to Oxygen Evolution’, J. Electrochem. Soc., vol. 142, no. 1, pp. 148–153,
Jan. 1995.
Bibliography 95
[76] Q. Han, ‘Hydrogen evolution reaction on amorphous Ni-S-Co alloy in alkaline

medium’, International Journal of Hydrogen Energy, vol. 28, no. 12, pp. 1345–1352,
2003.
[77] W. Hu, ‘Electrocatalytic properties of Ti~2Ni Ni Mo composite electrodes for
hydrogen evolution reaction’, International Journal of Hydrogen Energy, vol. 23, no.
4, pp. 253–258, 1998.
[78] R. Solmaz and G. Kardaş, ‘Fabrication and characterization of NiCoZn–M (M: Ag,
Pd and Pt) electrocatalysts as cathode materials for electrochemical hydrogen
production’, International Journal of Hydrogen Energy, vol. 36, no. 19, pp. 12079–
12087, Sep. 2011.
[79] A. O. Yüce, A. Döner, and G. Kardaş, ‘NiMn composite electrodes as cathode
material for hydrogen evolution reaction in alkaline solution’, International Journal
of Hydrogen Energy, vol. 38, no. 11, pp. 4466–4473, Apr. 2013.
[80] T. Otagawa, ‘Lanthanum Nickelate as Electrocatalyst - Oxygen Evolution’, Journal
of the Electrochemical Society, vol. 129, no. 10, pp. 2391–2392, 1982.
[81] M. Hamdani, M. I. S. Pereira, J. Douch, A. Ait Addi, Y. Berghoute, and M. H.
Mendonça, ‘Physicochemical and electrocatalytic properties of Li-Co3O4 anodes
prepared by chemical spray pyrolysis for application in alkaline water electrolysis’,
Electrochimica Acta, vol. 49, no. 9–10, pp. 1555–1563, Apr. 2004.
[82] R. N. Singh, D. Mishra, Anindita, A. S. K. Sinha, and A. Singh, ‘Novel
electrocatalysts for generating oxygen from alkaline water electrolysis’,
Electrochemistry Communications, vol. 9, no. 6, pp. 1369–1373, Jun. 2007.
[83] M. Eldeab, M. Awad, A. Mohammad, and T. Ohsaka, ‘Enhanced water electrolysis:
Electrocatalytic generation of oxygen gas at manganese oxide nanorods modified
electrodes’, Electrochemistry Communications, vol. 9, no. 8, pp. 2082–2087, Aug.
2007.
[84] H.-C. Chien, W.-Y. Cheng, Y.-H. Wang, T.-Y. Wei, and S.-Y. Lu, ‘Ultralow
overpotentials for oxygen evolution reactions achieved by nickel cobaltite aerogels’,
Journal of Materials Chemistry, vol. 21, no. 45, p. 18180, 2011.
[85] X. Li, F. C. Walsh, and D. Pletcher, ‘Nickel based electrocatalysts for oxygen
evolution in high current density, alkaline water electrolysers’, Physical Chemistry
Chemical Physics, vol. 13, no. 3, p. 1162, 2011.
[86] H. Vandenborre, R. Leysen, H. Nackaerts, and P. Van Asbroeck, ‘A survey of five
year intensive R&D work in Belgium on advanced alkaline water electrolysis’,
International Journal of Hydrogen Energy, vol. 9, no. 4, pp. 277–284, 1984.
[87] G. Schiller, R. Henne, P. Mohr, and V. Peinecke, ‘High performance electrodes for an
advanced intermittently operated 10-kW alkaline water electrolyzer’, Int. J. Hydrog.
Energy, vol. 23, no. 9, pp. 761–765, Sep. 1998.
[88] D. M. See and R. E. White, ‘Temperature and Concentration Dependence of the
Specific Conductivity of Concentrated Solutions of Potassium Hydroxide’, J. Chem.
Eng. Data, vol. 42, no. 6, pp. 1266–1268, Nov. 1997.
[89] R. J. Gilliam, J. W. Graydon, D. W. Kirk, and S. J. Thorpe, ‘A review of specific
conductivities of potassium hydroxide solutions for various concentrations and
temperatures’, International Journal of Hydrogen Energy, vol. 32, no. 3, pp. 359–
364, Mar. 2007.
[90] F. Allebrod, C. Chatzichristodoulou, P. L. Mollerup, and M. B. Mogensen, ‘Electrical
conductivity measurements of aqueous and immobilized potassium hydroxide’,
International Journal of Hydrogen Energy, vol. 37, no. 21, pp. 16505–16514, Nov.
2012.
[91] S. S. Zumdahl, in Chemical principles, 4th ed., 2002, pp. 225–244.
[92] H. Matsushima, D. Kiuchi, Y. Fukunaka, and K. Kuribayashi, ‘Single bubble growth
during water electrolysis under microgravity’, Electrochemistry Communications,
vol. 11, no. 8, pp. 1721–1723, Aug. 2009.
96
[93] H. Matsushima, Y. Fukunaka, and K. Kuribayashi, ‘Water electrolysis under

microgravity: Part II. Description of gas bubble evolution phenomena’,
Electrochimica Acta, vol. 51, no. 20, pp. 4190–4198, May 2006.
[94] J. A. Koza, S. Mühlenhoff, P. Zabinski, P. A. Nikrityuk, K. Eckert, M. Uhlemann, A.
Gebert, T. Weier, L. Schultz, and S. Odenbach, ‘Hydrogen evolution under the
influence of a magnetic field’, Electrochimica Acta, vol. 56, no. 6, pp. 2665–2675,
Feb. 2011.
[95] S. H. Ahn, I. Choi, H.-Y. Park, S. J. Hwang, S. J. Yoo, E. Cho, H.-J. Kim, D.
Henkensmeier, S. W. Nam, S.-K. Kim, and J. H. Jang, ‘Effect of morphology of
electrodeposited Ni catalysts on the behavior of bubbles generated during the oxygen
evolution reaction in alkaline water electrolysis’, Chemical Communications, vol. 49,
no. 81, p. 9323, 2013.
[96] C. W. M. P. Sillen, E. Barendrecht, L. J. J. Janssen, and S. J. D. van Stralen, ‘Gas
bubble behaviour during water electrolysis’, International Journal of Hydrogen
Energy, vol. 7, no. 7, pp. 577–587, 1982.
[97] M. Wang, Z. Wang, and Z. Guo, ‘Understanding of the intensified effect of super
gravity on hydrogen evolution reaction’, International Journal of Hydrogen Energy,
vol. 34, no. 13, pp. 5311–5317, Jul. 2009.
[98] S.-D. Li, C.-C. Wang, and C.-Y. Chen, ‘Water electrolysis in the presence of an
ultrasonic field’, Electrochimica Acta, vol. 54, no. 15, pp. 3877–3883, Jun. 2009.
[99] T. Iida, H. Matsushima, and Y. Fukunaka, ‘Water Electrolysis under a Magnetic
Field’, J. Electrochem. Soc., vol. 154, no. 8, pp. E112–E115, Jan. 2007.
[100] L. Lao, C. Ramshaw, and H. Yeung, ‘Process intensification: water electrolysis in a
centrifugal acceleration field’, J Appl Electrochem, vol. 41, no. 6, pp. 645–656, Jun.
2011.
[101] V. M. Nikolic, G. S. Tasic, A. D. Maksic, D. P. Saponjic, S. M. Miulovic, and M. P.
Marceta Kaninski, ‘Raising efficiency of hydrogen generation from alkaline water
electrolysis – Energy saving’, International Journal of Hydrogen Energy, vol. 35, no.
22, pp. 12369–12373, Nov. 2010.
[102] M. P. Marčeta Kaninski, A. D. Maksić, D. L. Stojić, and Š. S. Miljanić, ‘Ionic
activators in the electrolytic production of hydrogen—cost reduction-analysis of the
cathode’, Journal of Power Sources, vol. 131, no. 1–2, pp. 107–111, May 2004.
[103] A. D. Maksic, S. M. Miulovic, V. M. Nikolic, I. M. Perovic, and M. P. Marceta
Kaninski, ‘Energy consumption of the electrolytic hydrogen production using Ni–W
based activators—Part I’, Applied Catalysis A: General, vol. 405, no. 1–2, pp. 25–28,
Oct. 2011.
[104] G. S. Tasic, S. P. Maslovara, D. L. Zugic, A. D. Maksic, and M. P. Marceta Kaninski,
‘Characterization of the Ni–Mo catalyst formed in situ during hydrogen generation
from alkaline water electrolysis’, International Journal of Hydrogen Energy, vol. 36,
no. 18, pp. 11588–11595, Sep. 2011.
[105] J. Fischer, H. Hofmann, G. Luft, and H. Wendt, ‘Fundamental investigations and
electrochemical engineering aspects concerning an advanced concept for alkaline
water electrolysis’, AIChE Journal, vol. 26, no. 5, pp. 794–802, 1980.
[106] M. G. Nayar, P. Ragunathan, and S. K. Mitra, ‘Development and operation of a high
current density high pressure advanced electrolysis cell’, International Journal of
Hydrogen Energy, vol. 5, no. 1, pp. 65–74, 1980.
[107] M. H. Miles, G. Kissel, P. W. T. Lu, and S. Srinivasan, ‘Effect of Temperature on
Electrode Kinetic Parameters for Hydrogen and Oxygen Evolution Reactions on
Nickel Electrodes in Alkaline Solutions’, J. Electrochem. Soc., vol. 123, no. 3, pp.
332–336, Jan. 1976.
[108] J. C. Ganley, ‘High temperature and pressure alkaline electrolysis’, International
Journal of Hydrogen Energy, vol. 34, no. 9, pp. 3604–3611, May 2009.
Bibliography 97
[109] A. C. Ferreira, E. R. Gonzalez, E. A. Ticianelli, L. A. Avaca, and B. Matvienko, ‘The

effect of temperature on the water electrolysis reactions on nickel and nickel-based
codeposits’, J Appl Electrochem, vol. 18, no. 6, pp. 894–898, Nov. 1988.
[110] F. Allebrod, C. Chatzichristodoulou, and M. B. Mogensen, ‘Alkaline electrolysis cell
at high temperature and pressure of 250 °C and 42 bar’, Journal of Power Sources,
vol. 229, pp. 22–31, May 2013.
[111] J. Larminie and A. Dicks, Fuel cell systems explained. J. Wiley, 2003.
[112] H. S. Hong, U.-S. Chae, S.-T. Choo, and K. S. Lee, ‘Microstructure and electrical
conductivity of Ni/YSZ and NiO/YSZ composites for high-temperature electrolysis
prepared by mechanical alloying’, Journal of Power Sources, vol. 149, pp. 84–89,
Sep. 2005.
[113] A. Brisse, J. Schefold, and M. Zahid, ‘High temperature water electrolysis in solid
oxide cells’, International Journal of Hydrogen Energy, vol. 33, no. 20, pp. 5375–
5382, Oct. 2008.
[114] W. Dönitz and E. Erdle, ‘High-temperature electrolysis of water vapor—status of
development and perspectives for application’, International Journal of Hydrogen
Energy, vol. 10, no. 5, pp. 291–295, 1985.
[115] D. M. Soares, O. Teschke, and I. Torriani, ‘Hydride Effect on the Kinetics of the
Hydrogen Evolution Reaction on Nickel Cathodes in Alkaline Media’, J.
Electrochem. Soc., vol. 139, no. 1, pp. 98–105, Jan. 1992.
[116] J. Y. Huot and L. Brossard, ‘Time dependence of the hydrogen discharge at 70°C on
nickel cathodes’, International Journal of Hydrogen Energy, vol. 12, no. 12, pp. 821–
830, 1987.
[117] R. M. Abouatallah, D. W. Kirk, and J. W. Graydon, ‘Long-term electrolytic hydrogen
permeation in nickel and the effect of vanadium species addition’, Electrochimica
Acta, vol. 47, no. 15, pp. 2483–2494, Jun. 2002.
[118] R. . Abouatallah, D. . Kirk, S. . Thorpe, and J. . Graydon, ‘Reactivation of nickel
cathodes by dissolved vanadium species during hydrogen evolution in alkaline
media’, Electrochimica Acta, vol. 47, no. 4, pp. 613–621, Nov. 2001.
[119] A. E. Mauer, D. W. Kirk, and S. J. Thorpe, ‘The role of iron in the prevention of
nickel electrode deactivation in alkaline electrolysis’, Electrochimica Acta, vol. 52,
no. 11, pp. 3505–3509, Mar. 2007.
[120] ‘http://www.gamry.com/assets/Application-Notes/Check-the-Impedance-of-Your-
Reference-Electrode.pdf’. .
[121] ‘http://www.burkert.com/media/COM_Chemical_Resistance_Chart.pdf’. .
[122] ‘http://greenhydrogen.dk/Default.aspx?ID=551’. [Online]. Available:
http://greenhydrogen.dk/Default.aspx?ID=551. [Accessed: 22-Nov-2013].
[123] ‘http://www.climatebuildings.dk/h2college.php’. [Online]. Available:
http://www.climatebuildings.dk/h2college.php. [Accessed: 22-Nov-2013].
[124] R. Baboian, in Corrosion tests and standards : application and interpretation, 2005,
p. 26.
[125] H. J. Miao and D. L. Piron, ‘Composite-coating electrodes for hydrogen evolution
reaction’, Electrochimica Acta, vol. 38, no. 8, pp. 1079–1085, Jun. 1993.
[126] E. Endoh, H. Otouma, T. Morimoto, and Y. Oda, ‘New Raney nickel composite-
coated electrode for hydrogen evolution’, International Journal of Hydrogen Energy,
vol. 12, no. 7, pp. 473–479, 1987.
[127] E. Endoh, H. Otouma, and T. Morimoto, ‘Advanced low hydrogen overvoltage
cathode for chlor-alkali electrolysis cells’, International Journal of Hydrogen Energy,
vol. 13, no. 4, pp. 207–213, 1988.
[128] Y. Choquette, H. Ménard, and L. Brossard, ‘Hydrogen discharge on a Raney nickel
composite-coated electrode’, International Journal of Hydrogen Energy, vol. 14, no.
9, pp. 637–642, 1989.
98
[129] Y. Choquette, L. Brossard, A. Lasia, and H. Menard, ‘Study of the Kinetics of

Hydrogen Evolution Reaction on Raney Nickel Composite‐Coated Electrode by AC
Impedance Technique’, J. Electrochem. Soc., vol. 137, no. 6, pp. 1723–1730, Jan.
1990.
[130] C. Hitz and A. Lasia, ‘Experimental study and modeling of impedance of the her on
porous Ni electrodes’, Journal of Electroanalytical Chemistry, vol. 500, no. 1–2, pp.
213–222, Mar. 2001.
[131] L. Chen and A. Lasia, ‘Study of the Kinetics of Hydrogen Evolution Reaction on
Nickel‐Zinc Alloy Electrodes’, J. Electrochem. Soc., vol. 138, no. 11, pp. 3321–3328,
Jan. 1991.
[132] J. Balej, J. Divisek, H. Schmitz, and J. Mergel, ‘Preparation and properties of raney-
nickel electrodes on Ni-Zn base for H2 and O2 evolution from alkaline-solutions. Part
I: electrodeposition of Ni-Zn alloys from chloride solutions’, Journal of applied
electrochemistry, vol. 22, no. 8, pp. 705–710, Aug. 1992.
[133] D. Miousse, A. Lasia, and V. Borck, ‘Hydrogen evolution reaction on Ni-Al-Mo and
Ni-Al electrodes prepared by low pressure plasma spraying’, J Appl Electrochem, vol.
25, no. 6, pp. 592–602, Jun. 1995.
[134] G. Schiller, R. Henne, and V. Borck, ‘Vacuum plasma spraying of high-performance
electrodes for alkaline water electrolysis’, Journal of Thermal Spray Technology, vol.
4, no. 2, pp. 185–194, Jun. 1995.
[135] M. L. Bakker, D. J. Young, and M. S. Wainwright, ‘Selective leaching of NiAl3 and
Ni2Al3 intermetallics to form Raney nickels’, Journal of Materials Science, vol. 23,
no. 11, pp. 3921–3926, Nov. 1988.
[136] M. Janssen and G. Rieck, ‘Reaction diffusion and kirkendall-effect in nickel
aluminium system’, Transactions of the metallurgical society of AIME, vol. 239, no.
9, pp. 1372–1385, 1967.
[137] M. F. Singleton, J. L. Murray, and P. Nash, ‘Al-Ni (Aluminium-Nickel)’, in Binary
Alloy Phase Diagrams, 1st. edition., vol. 1, 2 vols., T. B. Massalski, H. Okamoto, P.
R. Subramanian, and L. Kacprzak, Eds. American Society for Metals, 1986, p. 142.
[138] R. F. Hunter and P. Hohn, ‘Caustic etching of aluminum with matte finish and low
waste capability’, 509104625-Feb-1992.
[139] H. L. Watts and D. W. Utley, ‘Volumetric Analysis of Sodium Aluminate Solutions’,
Anal. Chem., vol. 25, no. 6, pp. 864–867, Jun. 1953.
[140] Y. Li, Y. Zhang, C. Yang, and Y. Zhang, ‘Precipitating sandy aluminium hydroxide
from sodium aluminate solution by the neutralization of sodium bicarbonate’,
Hydrometallurgy, vol. 98, no. 1–2, pp. 52–57, Aug. 2009.
[141] J. R. Davies, Aluminium and Aluminium Alloys. ASM International, 1993.
[142] A. Kyriakopolous, M. Lynn, and R. Ghomashchi, ‘Reactive interaction of molten
aluminum and solid nickel’, Journal of Materials Science Letters, vol. 20, no. 18, pp.
1699–1701, Sep. 2001.
[143] V. I. Dybkov, ‘Interaction of iron-nickel alloys with liquid aluminium’, J Mater Sci,
vol. 28, no. 23, pp. 6371–6380, Dec. 1993.
[144] R. Blagoj and M. Jon, ‘Interaction of solid nickel with liquid mixture of aluminum
and nickel and formation of intermetallic phases’, Metalurgija, vol. 16, no. 3, pp.
177–185.
[145] H. Liu, M. Bouchard, and L. Zhang, ‘An experimental study of hydrogen solubility in
liquid aluminium’, Journal of Material Science, vol. 30, no. 17, pp. 4309–4315, Sep.
1995.
[146] X. R. 任晓, G. Chen, W. Z. 周文龙, C. Wu, and J. Zhang, ‘Formation and growth
kinetics of intermediate phases in Ni-Al diffusion couples’, J. Wuhan Univ. Technol.-
Mat. Sci. Edit., vol. 24, no. 5, pp. 787–790, Oct. 2009.
Bibliography 99
[147] T. Jiang, M. J. C. Brym, G. Dube, A. Lasia, and G. M. Brisard, ‘Studies on the

AlCl3/dimethylsulfone (DMSO2) electrolytes for the aluminum deposition
processes’, Surf. Coat. Technol., vol. 201, no. 14, pp. 6309–6317, Apr. 2007.
[148] J. Fransaer, E. Leunis, T. Hirato, and J. P. Celis, ‘Aluminium composite coatings
containing micrometre and nanometre-sized particles electroplated from a non-
aqueous electrolyte’, J. Appl. Electrochem., vol. 32, no. 2, pp. 123–128, Feb. 2002.
[149] M. Miyake, H. Motonami, S. Shiomi, and T. Hirato, ‘Electrodeposition of purified
aluminum coatings from dimethylsulfone–AlCl3 electrolytes with trimethylamine
hydrochloride’, Surface and Coatings Technology, vol. 206, no. 19–20, pp. 4225–
4229, May 2012.
[150] T. Jiang, M. J. C. Brym, G. Dube, A. Lasia, and G. M. Brisard, ‘Studies on the
AlCl3/dimethylsulfone (DMSO2) electrolytes for the aluminum deposition
processes’, Surf. Coat. Technol., vol. 201, no. 14, pp. 6309–6317, Apr. 2007.
[151] European Hydrogen Association. Where will the energy for hydrogen production
come from? - Status and Alternatives-. [Online] 2006
http://www.lbst.de/ressources/docs2007/EHA_WhereWillH2ComeFrom_2007.pdf
(Accessed: 21 November 2013)
[152] EUR-Lex. Communication from the commission to the European Parliament, the
council, the European Economic and social committee and the Regions. Energy 2020
A strategy for competitive, sustainable and secure energy [Online] November 2010
http://eur-
lex.europa.eu/LexUriServ/LexUriServ.do?uri=COM:2010:0639:FIN:EN:HTML
(Accessed: 2 August 2012)
[153] F. Mohseni, M. Magnusson, M. Görling, and P. Alvfors, ‘Biogas from renewable
electricity – Increasing a climate neutral fuel supply’, Appl. Energy, vol. 90, no. 1,
pp. 11–16, Feb. 2012.
[154] C. A. Floudas, J. A. Elia, and R. C. Baliban, ‘Hybrid and single feedstock energy
processes for liquid transportation fuels: A critical review’, Comput. Chem. Eng., vol.
41, pp. 24–51, Jun. 2012.
[155] M. F. Hordeski, Alternative Fuels: The Future of Hydrogen, Second Edition, 2nd ed.
The Fairmont Press, Inc., 2008.
Appended Papers
Appended Papers
I. Development of durable and efficient electrodes for large-scale

alkaline water electrolysis
II. Electrochemical investigation of surface area effects on PVD Al-Ni

as electrocatalyst for alkaline water electrolysis
III. Investigations of the diffusion mechanism of PVD Al and Ni couples

at 610°C
IV. Unveiling the secrets of the Standard Hydrogen Electrode - An

inspiration for the on-going development of hydrogen electrocatalyst
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 8 2 2 1 e8 2 3 1
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/he
Development of durable and efficient electrodes for

large-scale alkaline water electrolysis
Cecilı́a Kristı́n Kjartansdóttir a,*, Lars Pleth Nielsen b, Per Møller a

a
Department of Mechanical Engineering, Technical University of Denmark, Produktionstorvet, Building 425,
2800 Kgs. Lyngby, Denmark
b
Danish Technological Institute, Kongsvang Allé 29, 8000 Aarhus C, Denmark
article info abstract
Article history: A new type of electrodes for alkaline water electrolysis is produced by physical vapour
Received 31 January 2013 depositing (PVD) of aluminium onto a nickel substrate. The PVD Al/Ni is heat-treated to
Received in revised form facilitate alloy formation followed by a selective aluminium alkaline leaching. The ob-
26 March 2013 tained porous Ni surface is uniform and characterized by a unique interlayer adhesion,
Accepted 18 April 2013 which is critical for industrial application. IR-compensated polarisation curves prepared in
Available online 21 May 2013 a half-cell setup with 1 M KOH electrolyte at room temperature reveals that at least 400 mV
less potential is needed to decompose water into hydrogen and oxygen with the developed
Keywords: porous PVD Al/Ni electrodes as compared to solid nickel electrodes. High-resolution
Alkaline water electrolysis scanning electron microscope (HR-SEM) micrographs reveal Ni-electrode surfaces charac-
Hydrogen evolution reaction terized by a large surface area with pores down to a few nanometre sizes. Durability tests
High surface area porous Ni were carried out in a commercially produced bipolar electrolyser stack. The developed
Plasma vapour deposition electrodes showed stable behaviour under intermittent operation for over 9000 h indicating
AleNi Thermo-Chemical diffusion no serious deactivation in the density of active sites.
Aluminium leaching Copyright ª 2013, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights
Raney nickel reserved.
1. Introduction of increased production capacity and load management with

no greenhouse emissions. The hydrogen can subsequently be
According to the World of Energy Outlook 2010, made by the stored and used for producing electricity via fuel cells, com-
International Energy Agency, renewable energy is the answer bustion engines or gas turbines, whenever needed. New ideas
for a more secure, reliable and sustainable future. This implies for using hydrogen as a raw material for production of syn-
that a higher percentage of the energy will have to come from thetic fuels, such as methane by the Sabatier process [2],
fluctuating renewable energy sources such as wind, sun and liquid fuels by the FischereTropsch synthesis [3] or simply
water. Simultaneously, the ever increasing demand and pumping the hydrogen gas into the existing natural gas
stronger legalisations towards reducing the CO2 emission infrastructure, the “power-to-gas” idea will suddenly be a re-
worldwide [1] makes it necessary to develop new efficient al- ality when hydrogen becomes available in large quantities [4].
ternatives for energy conversion, energy storage and load A variety of water electrolysis systems have been proposed
management. Using the excess electrical power from renew- and constructed over the years. Alkaline electrolysis is the
able energy sources, e.g. wind, solar and wave technologies, to most mature commercial water electrolysis technology and
produce hydrogen via water electrolysis, offers the possibility offers the advantages of simplicity and is the current standard
* Corresponding author. Tel.: þ45 45252118.

E-mail address: ckkj@mek.dtu.dk (C.K. Kjartansdóttir).
0360-3199/$ e see front matter Copyright ª 2013, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijhydene.2013.04.101
8222 i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 8 2 2 1 e8 2 3 1
for large-scale water electrolysis systems [5] [6]. However, residual elements were determined to be below 0.1%. The
finding low cost electrode materials that are both efficient and nickel plates were cut to form circular specimens with a
durable is one of the remaining challenges for alkaline water diameter of 3 cm, intended for small-scale electrochemical
electrolysis (AWE) techniques. measurements and scanning electron microscope in-
Traditionally, the cathode material for AWE systems was vestigations. 16 nickel plates, designed for a commercial bi-
made of steel and the anode material from nickel or nickel- polar electrolysis stack, were prepared from the same type of
coated steel [7]. The pioneer work of Paul Sabatier (1912) and nickel. All nickel specimens to be Al PVD treated were
Murray Raney (1925) on hydrogenation catalysts [8] enabled cathodically degreased for 2 min prior to the PVD process. The
Justi and Winsel to discover the highly porous Raney nickel as Al PVD was done in a non-reactive DC-magnetron sputtering
an efficient hydrogen electrocatalyst in an alkaline media mode using a CC800/9 SinOx coating unit from CemeCon AG.
already in 1961 [9]. The Ni substrates were heated and etched in situ by Ar sput-
Raney nickel catalysts are generally prepared by rapidly tering prior to sputter-depositing aluminium to remove nickel
cooling a molten solution of nickel and aluminium. The oxide (NiO) from the surface. The circular specimens were
cooling procedure controls which AleNi phases are formed in coated on one side, whereas the specimens for the electrolysis
the solid. After solidification, the product is crushed into a fine stack were coated on both sides. The thickness of the
powder. The catalyst powder is subsequently activated by aluminium coating ranged from 20 to 40 mm. The Al PVD
selectively leaching aluminium from the AleNi alloy [10]. plated specimens were subsequently heat treated in an at-
Lattice vacancies formed when leaching causes large surface mospheric furnace for 24 h at 610 C followed by a selective
areas and high density of active sites due to a high density of aluminium leaching.
reactive lattice defects [11]. The activated catalyst therefore Two circular specimen were leached according to the
provides superior performance compared to unactivated non- following procedure; 2 h in 1% NaOH at room temperature,
porous nickel cathodes. 20 h in 10% NaOH at room temperature and 4 h in 30% NaOH at
Plasma spraying leachable precursor alloys (Al3Ni and/or 100 C, respectively. These specimens will be referred to as
Al3Ni2) onto a nickel or steel support is an alternative and PVD Al/Ni 1. Two circular specimens were leached in 30% KOH
common technique applied when producing supported Raney and 10% KNaC4H4O6*4H2O at 80 C with stirring for 24 h. These
nickel electrocatalysts for alkaline electrolysis [12e19]. Cold specimens will be referred to as PVD Al/Ni 2. The aluminium
rolling and hot dipping of aluminium combined with a leaching procedure for the 16 large electrodes was identical to
thermo-chemical diffusion process have also been proposed the leaching procedure for the PVD Al/Ni 2 specimens.
[8] [20]. However, the reproducibility and durability of the
Raney nickel electrodes are often deficient [7]. 2.2. Structural characterisation and composition
When utilizing atmospheric plasma spraying (APS) to form
supported Raney nickel electrodes, formation of the electrical For cross section characterisation and composition analyses,
resistive and brittle Al2O3 phase cannot be avoided. During four PVD Al/Ni specimens, one as plated, one from each
vacuum plasma spraying (VPS), no oxygen is available to react leaching method and one unleached, were cut into
with the aluminium, and the initial Raney nickel structure 1 1.5 cm2. Each specimen was hot-mounted in CloroFast
appears to be highly active towards the hydrogen evolution resin and grinded down to 4000 grit, subsequently polished
reaction (HER) [18]. Producing a VPS Raney nickel structure with 3 mm diamond and 0.04 mm SiO2 particles. JEOL JSM 5900
with interlayer adhesion that can withstand the harsh gas scanning electron microscope (SEM) was used for the cross
erosion during AWE is however challenging. section investigations and an integrated energy-dispersive X-
In this paper, we report studies on large-scale production ray spectroscopy from Oxford Instruments was used for
of electrodes suitable for commercially available alkaline elemental analysis. The surface structure and morphology of
water electrolysis stacks. The electrodes are produced by the PVD Al/Ni 2 electrodes were characterized by means of a
plasma vapour deposition (PVD) followed by a thermo- FEI Quanta 200 ESEM FEG scanning electron microscope.
chemical diffusion process and alkaline leaching. Structural
characterisation of the electrodes is performed by high- 2.3. Electrochemical measurements
resolution scanning electron microscope (HR-SEM). The elec-
trocatalytic activity of the developed electrodes is studied The catalytic activity of the developed electrodes was evalu-
with steady-state electrochemical measurements and cyclic ated by means of potentiodynamic polarisation curves. The
voltammetry. Durability tests are carried out in an industrial electrocatalytic active surface area of the electrodes was
scale-electrolysis stack. determined by the amount of Ni hydroxide formed during
cyclic voltammetry. The measurements were carried out
using Gamry Reference 3000 potentiostat/galvanostat and a
2. Experimental three-electrode electrochemical cell made of teflon. The
reference electrode was a Hg/HgO electrode from Radiometer
2.1. Preparation of electrodes and the counter electrode was made of pure nickel. The
electrolyte contained 1 M KOH and the measurements were
Commercially available nickel plates with a thickness of operated at 25 C. All the electrochemical measurements
0.5 mm were used as an electrode substrate. The purity of the were IR-compensation by means of current interruption
nickel plates were determined by optical emission spectros- technique. All following potentials are specified against the
copy, detecting 99% Ni, 0.25% Mn, 0.14% Fe and 0.11% Al. Other standard hydrogen electrode (SHE) potential.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 3 8 ( 2 0 1 3 ) 8 2 2 1 e8 2 3 1 8223
Prior to the cathodic polarisation measurements, the

electrodes were conditioned at 0.8 V for 30 min. Thereafter
the electrodes were kept at open circuit potential (OCP) for
5 min or until 0.01 mV/s stability was reached. Subsequently
cathodic sweep from 0.5 V up to 0.6 A current, at scan rate of
1 mV/s, was performed. The experimental procedure for the
anodic polarisation curves was as follows; conditioning at
0.7 V for 30 min, OCP for 5 min, or until stability of 0.01 mV/s
was reached and anodic sweep from 0.5 V up to 0.6 A current
density at 1 mV/s scan rate.
The cyclic voltammetry measurements were performed
from 0.9 to 1.1 V with upper limit of 0.1 V at the scan rate
of 50 mV/s. Prior the measurement the electrodes were
conditioned at 1.5 V for 30 min and 0.9 V for 5 min.
2.4. Adhesion and durability test Fig. 2 e Cross section SEM micrograph of a PVD Al/Ni
electrode as plated.
The 16 large PVD Al/Ni electrodes were mounted in a 17 cell
bipolar, non-zero gap electrolysis stack produced by Green-
diffusion couples at that temperature can be formed. The
Hydrogen.dk. The stack contained 30% KOH and was operated
diffusion coefficient for the AleNi system is used for selecting
under altered conditions, at a maximum temperature of 80 C
appropriate heat treatment parameters, they have been
and a pressure of 22 bars for more than 9000 h. As the elec-
calculated elsewhere [21]. The red horizontal line in the phase
trodes were treated with the alkaline-etched PVD Al/Ni on
diagram in Fig. 1 indicates which AleNi diffusion couples are
both sides, the developed surface functioned both as an anode
thermodynamically stable at 610 C and atmospheric pres-
and as a cathode in the electrolysis stack.
sure. The thickness of each intermetallic phase formed de-
The mechanical strength and adhesion properties of the
pends on the amount of Ni and Al available in the diffusion
developed electrodes were evaluated via a bending test com-
system and the heat treatment parameters.
parable to ASTM B571, 1997 (2008) e1.
The SEM micrographs in Figs. 2 and 3 show cross section
micrographs of a PVD Al/Ni electrode prior and after the
heat treatment. Comparing the energy-dispersive X-ray
3. Results and discussions
spectroscopy (EDS) data, Table 1, with the AleNi phase di-
agram, it is supposed that the three following AleNi inter-
3.1. Structural characterisation and composition
metallic phases are formed during heat treatment; Al3Ni2,
AlNi and AlNi3, seen from the top towards the pure Ni
When heat-treated at 610 C, a thermo-chemical diffusion
substrate. This is in agreement with the findings of Janssen
process takes place at the contact area between the
and Rieck [21]. The lowermost phase in Fig. 3 is the unaf-
aluminium and nickel phase. The aluminium atoms diffuse
fected nickel substrate. The majority of the intermetallic
into the nickel structure and thermodynamically stable
phases formed is the strong, and yet leachable, Al3Ni2 phase
[22].
Fig. 1 e Phase diagram for the AleNi diffusion couples

adopted from [23]. The horizontal line indicates the Fig. 3 e Cross section SEM micrograph of a PVD Al/Ni
thermo-chemical diffusion temperature (610 C) selected electrode after heat treatment at 610 C for 24 h, prior to
for the electrode development. leaching. The numbers refer to the EDS analysis in Table 1.
Table 1 e Results from the cross section EDS analysis on PVD Al/Ni electrodes before and after the first and the second
leaching procedure. The phase numbers refer to the numbers in Figs. 3e5. All elements from the periodic table are
analysed.
Phase nr. Name of phase Before leaching After the 1. leaching procedure After the 2. leaching procedure
Al wt.% Ni wt.% O wt% Al wt.% Ni wt.% O wt% Al wt.% Ni wt.%
1 Al3Ni2 37 63 4 21 75 7 13 80
2 Al3Ni2 37 63 36 64 4 15 81
3 AlNi 30 70 30 70 29 71
4 AlNi3 13 87 13 87 14 86
5 Ni 100 100 100
Fig. 4 and Fig. 5 show cross section SEM micrographs of the

PVD Al/Ni electrodes after heat treatment followed by the first
and second alkaline leaching procedure, PVD Al/Ni 1 and PVD
Al/Ni 2, respectively. Using the first leaching procedure, about
5 mm skeletal Al/Ni residue is formed. However, when selec-
tively leaching the Al with the second leaching procedure, the
Fig. 4 e Cross section SEM micrograph of a PVD Al/Ni 1

electrode after heat treatment at 610 C for 24 h and
aluminium leaching for 2 h in 1% NaOH at room
temperature, 20 h in 10% NaOH at room temperature and
4 h in 30% NaOH at 100 C, respectively.
Fig. 5 e Cross section SEM micrograph of a PVD Al/Ni 2

electrode after heat treatment at 610 C for 24 h and
selective leaching of aluminium in 30% KOH and 10% Fig. 6 e HR-SEM micrographs of a PVD Al/Ni 2 surface after
KNaC4H4O6*4H2O at 80 C for 24 h, respectively. heat treatment and complete alkaline leaching.
a b
Fig. 7 e Voltammograms recorded on polished Ni, PVD Al/Ni 1 and PVD Al/NI 2 electrodes from L0.9 to L1.1 V with upper
limit of L0.1 V at the scan rate of 50 mV/s, in 1 M KOH at 25 C the anodic peek represents the oxidation of the Ni surface to
a-Ni(OH)2.
entire Al3Ni2 phase is leached, resulting in considerably larger was expected to be formed during the heat treatment. How-
thickness of the porous nickel layer. The PVD Al/Ni 1 elec- ever, after the heat treatment, no Al3Ni phase was formed in
trodes also have larger amount of aluminium remaining in the the inter-diffusion structure. Therefore, the leaching pro-
leached structure, or 21 wt% compared to 13e15 wt% for the cedures at room temperature do not contribute to the leaching
PVD Al/Ni 2 electrodes. The difference can be due to the slow process. The 4 h of leaching at 100 C in NaOH is evidently not
leaching kinetics of the Al3Ni2 phase below 70 C in alkaline enough for complete leaching of the Al3Ni2 structure. In
media [22]. For the first leaching procedure, the leaching steps addition, some of the Al could possibly have been redeposited
at room temperature were selected for thorough leaching of into the pores during the first leaching method since no KNa-
the Al3Ni phase, which, according to the AleNi phase diagram, tartrate tetrahydrate was used in that process [24]. The reason
Table 2 e Results from cyclic voltammetry investigations at 1 M KOH and 25 C.

Specimen Geometrical Charge associating The active electrochemical Roughness
surface area (cm2) with the formation of a-Ni(OH)2 (mC) surface area (cm2)a factorb
Polished Nickel 2 1.1 2.1 1,05

PVD Al/Ni 1 2 246.3 479.2 239.6
PVD Al/Ni2 2 558.2 1086.0 543.0
a The charge associating with formation of a-Ni(OH)2 recorded divided by the charge associating with the formation and reduction of a
monolayer of a-Ni(OH)2 (514 C cm2) for nickel electrode according to [28].
b The active electrochemical surface area divided by the geometrical surface area.
for adding KNa-tartrate tetrahydrate to the alkaline leaching

solutions is that it works as a complexing agent for the
leached aluminium according to [25]:
AlðOHÞ 2 2
4 þ nðC4 H4 O6 Þ /AlðOHÞ3 ðC4 H4 O6 Þn þ OH

hereby preventing aluminium hydroxide precipitates to

redeposit into the nickel pores [26]:
AlðOHÞ
4 /AlðOHÞ3 þ OH

The cross section images reveal a highly porous nickel

microstructure with pores in the range of approximately
0.5e1.5 mm. For nanostructure investigations, high-resolution
SEM studies are required. HR-SEM micrographs of the surface
of a PVD Al/Ni 2 specimen at different magnifications are
shown in Fig. 6. When looking at the HR-SEM micrographs,
nanopores down to at least 20 nm are revealed.
Fig. 8 e Cathodic polarisation curves recorded on PVD
Al/Ni 1, a PVD Al/Ni 2 and a polished nickel (99%), repeated
3.2. Electrochemical measurements
three times, in 1 M KOH electrolyte at 25 C.
The anodic peaks in the voltammograms in Fig. 7 represents

the oxidation of the Ni surface to a-Ni(OH)2 [27]. The charge
associating with the formation and reduction of a mono-
times in a row. As seen from the figure, the measurements are
layer of a-Ni(OH)2 on plain nickel is known to be 514 mC cm2
consistent. From the polarisation curves it is evident that the
[28]. The ratio between the charge associated with the for-
PVD Al/Ni 1 is more active towards the HER compared to the
mation of a-Ni(OH)2 on the tested surfaces and the known
polished nickel but less active compared to the PVD Al/Ni 2
charge value for nickel can be used to estimate the actual
electrodes. At 200 mA/cm2 (which is a typical operation cur-
electrochemical surface area. The roughness factor of the
rent density for commercial electrolysers) the PVD Al/Ni 1
surfaces can then be estimated by dividing the actual elec-
electrode has about 280 mV less hydrogen overpotential
trochemical surface area by the geometrical surface area.
compared to polished Ni. The PVD Al/Ni 2 electrode has
The charge associating with the formation of a monolayer of
however about 360 mV less overpotential, compared to the
a-Ni(OH)2 on polished Ni, PVD Al/Ni 1 and PVD Al/Ni 2 are
nickel. At higher current density values the overpotential of
found to be 1.1 mC, 246.3 mC and 558.2 mC, respectively.
the PVD Al/Ni 1 structure degreases more rapidly than for the
Due to irreversible aging of a-Ni(OH)2 to b-Ni(OH)2 under the
other two structures. This difference can be due to the large
voltammetry conditions the first cycle from the cycle sweeps
amount of aluminium (about 21 wt%) that is remaining in the
was chosen for peek estimation. The results from the vol-
top layer of the electrode.
tammogram measurements are collected in Table 2. From
In order to minimize the complexity and cost for the bi-
the roughness factors it is evident that the electrodes
polar electrode production, developing one type of
leached with the first leaching method (PVD Al/Ni 1) have
about 230 times larger electrochemical active surface area
compared to the polished nickel electrode. The electrodes
leached with the second leaching method (PVD Al/Ni 2) have
about 517 times larger electrochemical active surface area
compared to polished nickel and 2 times larger electro-
chemical active surface area compared to PVD Al/Ni 1. The
cyclic voltammetry analysis support that the developed
electrodes are highly porous as indicated by the SEM mi-
crographs. The cyclic voltammograms do however not say
anything about how deep porous layer contributes to the
electrochemical active surface. It is indeed expected that
only few micrometres of the porous layer contributes to the
actual electrochemical surface area. The large difference in
active surface area between the electrodes produced with
the two different etching methods is supposed to be due to
the large amount of aluminium that is remaining in the top
structure of the PVD Al/Ni 1 electrodes compare to the PVD
Al/Ni 2 electrodes. That results in less pores and smaller
surface area at the surface. Fig. 9 e Anodic polarisation curves recorded on PVD
Fig. 8 shows cathodic polarisation curves recorded on a Al/Ni 1, a PVD Al/Ni 2 and a polished nickel (99%), repeated
PVD Al/Ni 1, PVD Al/Ni 2 and polished nickel repeated three three times, in 1 M KOH electrolyte at 25 C.
Fig. 10 e Pourbiax diagram for the AleNieH2O system at 25 C.
electrocatalytic surface that is efficient and durable for both of the nanopores with oxides already after a short time of
the anodic and cathodic reaction will be of major advantage. operation. This will decrease the actual surface area of the
Although, the developed electrocatalyst was mainly designed electrode, decaying the performance towards that of a smooth
for the HER process, testing of its performance for the oxygen nickel surface. The following anodic half-cell reactions and
evolution reaction (OER) is also of interest. dissolution processes are suggested to describe the mentioned
Fig. 9 shows anodic polarisation curves recorded on a PVD decomposition reactions of the anode in a strong alkaline
Al/Ni 1, PVD Al/Ni 2 and a polished nickel surfaces. As seen solution:
from the figure, PVD Al/Ni 1 and PVD Al/Ni 2 are similarly
active towards OER at current densities up to 300 mA/cm2.
Ni þ 3OH /NiðOHÞ3 þ 2e
Only 50 mV is gained by using PVD Al/Ni 2 for the OER
compared to polished nickel. As seen from the cyclic vol-
Ni þ 2OH /NiO þ H2 O þ 2e
tammetry measurements and SEM images, the surface area of
the developed electrocatalysts are much larger than the sur-
NiO þ H2 O þ OH /NiðOHÞ
face area of the smooth electrode, thus more difference in 3
efficiency would be expected. However, according to the It is noteworthy that the slopes for the PVD Al/Ni 2 elec-
Pourbaix-diagram in Fig. 9, nickel oxides are formed during trodes and the nickel electrodes are parallel for both HER and
the OER at the electrode surfaces. It is therefore probable that OER, see Figs. 8 and 9. This indicates that the same reaction
the skeletal nickel structure will be transformed into nickel
oxide or corrode. The result of this might be blockage of some
Table 3 e The chemical reactions and corresponding

Gibbs free energy for alkaline aluminium leaching of the
AleNi intermetallic phases in the Pourbaix diagram in
Fig. 10. The energy values are calculated at 20 C.
Chemical reaction Gibbs free
energy [DG0 ]
Al3Ni þ 3OH þ 9H2O / 3Al(OH)4 þ 4.5H2 þ Ni 1157.3 kJ

Al3Ni2 þ 3OH þ 9H2O / 3Al(OH)4 þ 4.5H2 þ 2Ni 1026.7 kJ
AlNi3 þ OH þ 3H2O /Al(OH)4 þ 1.5H2 þ 3Ni 284.4 kJ
Fig. 11 e Single bend test on a heat-treated and leached
AlNi þ OH þ 3H2O / Al(OH)4 þ 1.5H2 þ Ni 318.4 kJ
PVD Al/Ni 2 electrode.
Fig. 12 e The bipolar construction of the electrolysis stack used for durability testing.
mechanism is taking place and the difference in activity mechanical stability of the electrodes during operation.
comes from the difference in the actual surface area and the Fig. 10 shows a Pourbaix diagram calculated for the
increase in the electrochemical active sides. That is, the AleNieH2O system. The two dotted lines indicate the area
intrinsic electrocatalytic properties are evidently the same or where water is thermodynamically stable. Thus, the HER
similar for the developed electrodes and pure nickel. For the reaction can only take place below these lines, whereas the
PVD Al/Ni 1 structure where there is more aluminium OER can only take place above the lines. The diagram shows
remaining in the structure this is not the chase. The polar- that none of the AleNi intermetallic phases produced during
isation curves therefore show clearly that proper etching of the heat treatment is thermodynamically stable in strong
the developed electrodes is crucial for attaining the optimal alkali at the operating potentials. The chemical reactions and
efficiency, especially at higher current densities. Gibbs free energy for aluminium leaching of the intermetallic
phases in the Pourbaix diagram are shown in Table 3
3.3. Adhesion and durability
The durability of electrodes for AWE depends on the corro-

sion resistivity of the material in the operating media and the
Fig. 14 e Current vs. voltage and efficiency data captured

on the 17-cell bipolar electrolysis stack after more than
Fig. 13 e Current vs. voltage data captured on the 17-cell 9000 operating hours. The stak was operated at 60 C and
bipolar electrolysis stack during the first month of operation. 22 bars. The efficiency calculations are based upon the
The operation conditions were 40e45 C and at 10 bars. value of the thermal neutral voltage (E [ 1.48 V).
sufficient, the electrolyte will eventually penetrate in-

between the two layers and electrochemical reactions will
take place leading to gas formation in the interphase. When
this happens, the gas bubbles apply mechanical forces on the
relatively brittle porous microstructure leading to gas erosion
corrosion. The catalytic active surface will in that case scale
off and the electrodes will lose their high active catalytic
behaviour and the efficiency of the electrolysis stack will
decrease.
For testing the adhesion between the developed electro-
catalyst and the nickel substrate, bending tests were con-
structed on the leached PVD Al/Ni 2 electrodes produced.
Fig. 11 shows a reproducible image of a tested electrode after
bending. As seen from the figure, the PVD Al/Ni 2 coating
does not peel off during the bending. This is an indication
of good adhesion between the substrate and the
Fig. 15 e Efficiency measurement captured on the 17-cell
electrocatalyst.
bipolar electrolysis stack at 22 bars and 200 mA/cm2
Durability test on sixteen large electrodes containing the
current density, after more than 9000 operating hours. The
developed electrocatalytic surface was carried out in a 17 cell
efficiency calculations are based upon the value of the
bipolar electrolyser stack produced by the company Green-
thermal neutral voltage (E [ 1.48 V).
Hydrogen.dk. The end electrodes in the stack were made of
pure nickel. Fig. 12 shows the bipolar constructions of the
stack.
Although the thermodynamics indicate otherwise, it is As mentioned before, the stack was operated under altered
evident from the EDS analyses on the leached structures in conditions, maximum temperature of 80 C and a pressure of
Table 1 that the AlNi and AlNi3 phases are stable in strong 22 bars, for more than 9000 h. For safety reasons the stack was
alkali. It can therefore be assumed that the leaching kinetics operated under low temperature and pressure in the begin-
of AlNi and AlNi3 at elevated temperatures are too slow for the ning of the testing period.
mechanism to take place. In addition, according to the ther- Figs. 13e15 show the current vs. voltage and temperature
modynamics, nickel is not stable in strong alkali inside the vs. voltage data from the 17-cell bipolar electrolysis stack
water-stable area. If this was the case, the Ni electrocatalysts from the first month of the operating period and after over
would decompose when the electrolysis stack was switched- 9000 h operation. The fact that the test electrolysis stack was
off. However, due to the use of nickel as an anode in AWE operated under different conditions in the beginning and the
for many years, it is evident that the dissolution of Ni in end of the testing period makes durability comparison chal-
alkaline media is slow, making nickel one of the most lenging but not impossible. In Table 4 three values from the
corrosion-resistant metals of non-noble metals in that media. stack measurements showed in Figs. 13 and 15 have been
During commercial alkaline water electrolysis, the high chosen for durability assessment, one from the beginning of
pH of the electrolyte, intermediate temperatures, and gas the test period and two from after 9000 h. These values
evolution all contribute in making the electrode environment indicate that no serious deactivation have taken place on the
extremely harsh. The knowledge of the chemical resistivity electrodes during the 9000 h testing period. It should be
of nickel in strong alkaline solution has made nickel the mentioned that these efficiency values are lower than the
material of choice when developing electrodes for AWE. actual cell efficiencies. Designing an electrolysis stack with
However, the large mechanical stresses caused by gas evo- low energy losses is a complex matter. Large losses can come
lution during the process are often neglected. Adhesion be- from stray current between cells and piling op of gases due to
tween the electrocatalyst and the substrate is crucial for the improper stack design [29]. This is however out of the scope
mechanical resistivity of the electrode. If the adhesion is not of the present work.
Table 4 e Selected data from Figs. 11e13 for durability assessment.

Temperature ( C) Pressure Stack Cell current Stack Cell Stack
(Bar) current (A) density (A/cm2) voltage (V) voltage (V) efficiencya (%)
First month 40 10 40 0.15 33 1.94 76

of the operation period
After more 40 22 54 0.2 35 2.05 73
than 9000 h operation
After more 60 22 40 0.15 32 1.88 79
than 9000 h operation
a The efficiency calculations is based upon the value of the thermal neutral voltage (E ¼ 1.48 V).
Communities. Journal of Applied Electrochemistry

4. Conclusions 1988;18(1):1e14.
[6] Zeng K, Zhang D. Recent progress in alkaline water
The development of the present electrodes has been caried electrolysis for hydrogen production and applications.
out with the focus on durability, stability and potential large- Progress in Energy and Combustion Science
2010;36(3):307e26.
scale production. The electrocatalytical activity of an elec-
[7] Lessing PA. Materials for water electrolysis cells. In:
trode depends on the electron configuration and the structure
Jones RH, Thomas GJ, editors. Materials for the
and geometry of the catalyst. In the well-known volcano plot, hydrogen economy. 1st ed. Boca Raton: CRC Press; 2007.
where the metal-hydrogen bond strength of a material is p. 38e41.
plotted against the exchange current density of the HER [30], it [8] Method of producing finely-divided nickel. U.S. Patent
is shown that Ni is the most active pure metal among the non- 162819010-May-1927.
noble candidates. Finding electrocatalytic material with good [9] Chen E. History. In: Hoogers G, editor. Fuel cell technology
Handbook. USA: CRC Press; 2002.
intrinsic properties and increasing the efficiency of the cata-
[10] Wainwright MS. Skeletal metal catalysts. In: Ertl G,
lyst by enlarging the actual surface area have been described Knözinger H, Weitkamp J, editors. Preparation of solid
by various authors [31e35]. However, information about the catalysts. Wiley-VCH Verlag GmbH; 1999. p. 28e43.
mechanical stability and durability of the modified electro- [11] Bagotsky VS. Fuel cells: problems and solutions. John Wiley
catalysts is often lacking. & Sons; 2012209e10.
The present electrodes show good potentials as electrodes [12] Sillitto SMA, Adkins NJE, Ormerod RM, Paul E,
for large-scale alkaline water electrolysis systems. The devel- Hodgson DR. Wragg AA, editor. Characterisation of
advanced Raney nickel electrocatalytic coatings produced
oped electrodes have shown to be highly efficient for the
by the direct spraying method. Rugby: Inst Chemical
hydrogen evolutions reaction, however, more efficiency is Engineers; 1999.
desired for the oxygen evolutions reaction. One of the major [13] Sillitto SMA, Adkins NJE, Hodgson DR, Paul E, Ormerod RM.
drawbacks of using nickel for the anodic process is that a low Electrochemical testing and structural characterisation of
electrical conductive oxide layer is formed during electrolysis. nickel based catalytic coatings produced by direct spraying.
Furthermore, the nanopores in the developed structure will In: Lednor PW, Nagaki DA, Thompson LT, editors. Advanced
catalytic materials-1998. Warrendale: Materials Research
presumably be filled up with oxides and will therefore not
Society; 1999. p. 23e9.
contribute to the actual surface area of the electrode. The next
[14] Kellenberger A, Vaszilcsin N, Brandl W, Duteanu N. Kinetics
step in the development process is to optimize the anode ma- of hydrogen evolution reaction on skeleton nickel and
terial and carry out further durability testing. nickel-titanium electrodes obtained by thermal arc spraying
technique. International Journal of Hydrogen Energy
2007;32(15):3258e65.
[15] Kellenberger A, Vaszilcsin N, Brandl W. Roughness factor
Acknowledgement evaluation of thermal arc sprayed skeleton nickel
electrodes. Journal of Solid State Electrochemistry
The authors wish to thank Jørgen Jensen and Alexander 2007;11(1):84e9.
[16] Birry L, Lasia A. Studies of the hydrogen evolution reaction
Dierking from the company GreenHydrogen.dk for their help
on Raney nickel-molybdenum electrodes. Journal of Applied
with the durability tests. The authors also want to thank Sune Electrochemistry 2004;34(7):735e49.
Egelund, Melanie Röefzaad and Michael Caspersen for their [17] Kellenberger A, Vaszilcsin N. The determination of the
help with the electrochemical measurements. Financial sup- roughness factor of skeleton nickel electrodes by cyclic
port from The Energy Technology Development and Demon- voltammetry. International Journal of Hydrogen Energy
stration Program in Denmark (EUDP) (project number: 2007;32(15):3258e65.
[18] Schiller G, Henne R, Borck V. Vacuum plasma spraying of
63011-0200) is also gratefully acknowledged.
high-performance electrodes for alkaline water
electrolysis. Journal of Thermal Spray Technology
1995;4(2):185e94.
references [19] Fournier J, Miousse D, Legoux JG. Wire-arc sprayed nickel
based coating for hydrogen evolution reaction in alkaline
solutions. International Journal of Hydrogen Energy
[1] Energy 2020-Strategy of the European Commission for a 1999;24(6):519e28.
Competitive, sustainable and secure power Engineering. [20] Borucin ski T, Rausch S, Wendt H. Raney nickel activated
Energetika Prague 2011;61(Part 2):159e67. H2-cathodes part II: correlation of morphology and
[2]. Mohseni F, Magnusson M, Görling M, Alvfors P. Biogas from effective catalytic activity of Raney-nickel coated
renewable electricity e increasing a climate neutral fuel cathodes. Journal of Applied Electrochemistry
supply. Applied Energy 2012;90(1):11e6. 1992;22(11):1031e8.
[3] Floudas CA, Elia JA, Baliban RC. Hybrid and single feedstock [21] Jansssen M, Rieck G. Reaction diffusion and kirkendal-
energy processes for liquid transportation fuels: a critical effect in nickel aluminium system. Transactions of the
review. Computers & Chemical Engineering 2012;41:24e51. Metallurgical Society of AIME 1967;239(9):1372e85.
[4] Hordeski MF. Alternative fuels: the future of hydrogen. 2nd [22] Bakker ML, Young DJ, Wainwright MS. Selective leaching of
ed. Lilburn: The Fairmont Press, Inc; 2008. NiAl3 and Ni2Al3 intermetallics to form Raney nickels.
[5] Wendt H, Imarisio G. Nine years of research and Journal of Materials Science 1988;23(11):3921e6.
development on advanced water electrolysis. A review of the [23] Singleton MF, Murray JL, Nash P. Al-Ni (Aluminium-Nickel).
research programme of the Commission of the European In: Massalski TB, Okamoto H, Subramanian PR, Kacprzak L,
editors. Binary alloy phase diagrams. 1st. ed. Ohio: American [30] Trasatti S. Electrochemical theory j hydrogen evolution. In:
Society for Metals; 1986. p. 142. Garche J, editor. Encyclopedia of electrochemical power
[24] Hunter RF, Hohn P. Caustic etching of aluminum with matte sources. Amsterdam: Elsevier; 2009. p. 41e8.
finish and low waste capability. U.S. Patent 509104625-Feb- [31] Eldeab M, Awad M, Mohammad A, Ohsaka T. Enhanced
1992. water electrolysis: electrocatalytic generation of oxygen
[25] Watts HL, Utley DW. Volumetric analysis of sodium gas at manganese oxide nanorods modified electrodes.
aluminate solutions. Analytical Chemistry Electrochemistry Communications 2007;9(8):2082e7.
1953;25(6):864e7. [32] Brennecke PW, Ewe HH. Hydrogen evolution of highly
[26] Li Y, Zhang Y, Yang C, Zhang Y. Precipitating sandy porous Raney nickel cathodes in alkaline electrolyte.
aluminium hydroxide from sodium aluminate solution by Energy Conversion and Management 1991;31(6):585e94.
the neutralization of sodium bicarbonate. Hydrometallurgy [33] Dyer CK. Improved nickel anodes for industrial water
2009;98(1e2):52e7. electrolyzers. Journal of the Electrochemical Society
[27] Correia AN, Machado SAS. Hydrogen evolution on 1985;132(1):64e7.
electrodeposited Ni and Hg ultramicroelectrodes. [34] Martı́nez WM, Fernández AM, Cano U, Sandoval JA.
Electrochimica Acta 1998;43(3e4):367e73. Synthesis of nickel-based skeletal catalyst for an alkaline
[28] Brown IJ, Sotiris S. Electrodeposition of Ni from a high electrolyzer. International Journal of Hydrogen Energy
internal phase emulsion (HIPE) template. Electrochimica 2010;35(16):8457e62.
Acta 2001;46(17):2711e20. [35] Herraiz-Cardona I, Ortega E, Pérez-Herranz V. Impedance
[29] Caspersen M, Kirkegaard JB. Modelling electrolyte study of hydrogen evolution on Ni/Zn and NieCo/Zn
conductivity in a water electrolyzer cell. International stainless steel based electrodeposits. Electrochimica Acta
Journal of Hydrogen Energy 2012;37(9):7436e41. 2011;56(3):1308e15.
Appended Paper II 1
Electrochemical investigation of surface area effects on PVD Al‐Ni as

electrocatalyst for alkaline water electrolysis
Cecilía Kjartansdóttira, Michael Caspersenb, Sune Egelundb and Per Møllera.
a Department of Mechanical Engineering, The Technical University of Denmark
b
Siemens A/S
Abstract
A thermo-chemical diffusion process on about 20 µm physical vapour deposited aluminium onto a nickel
substrate, leads to a rapid formation of Al/Ni intermetallic layer that is particularly acceptable for dissolution
of aluminium in strong alkali. The geometry and the structure of the final skeletal nickel coatings can be
manipulated by altering the time interval of the diffusion. In that way the actual electrochemical surface area
and, thus, the electrocatalytic activity of the coatings towards HER and OER can be influenced. Cyclic
voltammetry (CV) and electrochemical impedance spectroscopy (EIS) investigations, display that the
electrocatalytic surface area increases in proportion to the thickness of the developed porous nickel structure.
In the case of the HER an increase in roughness factor (Rf) from 1 (for polished Ni) to 800, results in
reduction of the hydrogen overpotential of 337 mV. When further increasing Rf up to above 2000, additional
40 mV are gained. For the OER, smaller roughness values were detected with the same activity trend as for
the HER. The electrocatalyst are however found not to be stable in the anodic environment during
electrolysis. The corrosion mechanism of a skeletal nickel electrocatalyst during the OER in an alkaline
environment is briefly discussed.
The structure and composition of the electrocatalysts are characterised by scanning electron microscopy and
X-ray diffraction. The actual electrocatalytic surface area and the electrocatalytic behaviour are studied with
potentiodynamic polarisation, CVs and EISs.
1 Introduction
Nickel is known for its good corrosion resistivity in strong alkali [1][2] and its relatively good oxygen and
hydrogen electrocatalytic properties [3]. Nickel is therefore typically the core material used in bipolar
electrodes for alkaline water electrolysis systems [4]. In 1961 Justi and Winsel [5] discovered that Raney
nickel (originally developed as a catalyst for hydrogenation of vegetable oils) was an effective hydrogen
electrocatalyst for alkaline electrolysis. The principle behind the Raney nickel catalysts is that Al or Zn is
selectively leached from NiAl or NiZn alloys [6–8]. Lattice vacancies formed when leaching result in large
surface area and high density of lattice defects, which again leads to formation of additional active sides for
the electrocatalytic reaction to take place [9]. Later, many authors have published an increase in
electrocatalytic activity towards the hydrogen evolution reaction (HER) by selectively leaching one or more
elements from metal alloys [10–18]. One recent example is the work of Birry and Lasia [14] where Al-Ni
and Al-Ni-Mo alloys were prepared by the means of pressing and heating and by vacuum plasma spraying.
They reported HER overpotentials for alkaline leached Al3Ni and Al3Ni2 alloys measured in 1 M KOH at
25C to be 136 and 280 mV, respectively. They also showed that drastic reduction in overpotential could be
reached by adding Mo to high aluminium containing alloys. HER overpotential of only 67 mV was measured
for a plasma sprayed and alkaline leached NiAl5.95Mo0.66 alloy.
2
Raney nickel has also, occasionally, been proposed as a suitable electrocatalyst for the oxygen evolution
reaction (OER). In almost all cases it has, independently of preparation procedure, proved to be superior to
elemental nickel [19,20]. In [16] Raney nickel electrodes were prepared by plasma spraying with a mixture
of Ni-Al alloy powder and Co3O4 particles. This procedure provided a catalyst with an oxygen overpotential
reduction of approximately 150 mV when compared to plain nickel. The stability of this catalyst was later
found to be acceptable when the electrodes were subjected to test in a full scale electrolyser [21]. Maunowski
and Jtilich [22] reported results where Raney nickel was shown to be superior as anode catalyst when
compared to other well-known and active catalysts such as NiCo2O4, Co3O4, LaNiO3 and La0.5Sr0.5CoO3.
In a previous study [23], porous nickel electrodes were produced by physical vapour deposition (PVD) of Al
onto a nickel substrate followed by thermo-chemical diffusion and alkaline aluminium leaching. The
electrodes were shown to be durable and have good cell potentials as bipolar electrodes for industrial scale
alkaline water electrolysers. However, the optimal thickness of the electrocatalyst had not been studied.
When producing porous nickel electrocatalyst in that manner, there is a trade-off between the thickness of
the electrocatalyst and the build-up of internal stresses in the coating. During manufacturing, both the
interdiffusion and the leaching procedures facilitate internal stresses which escalate in proportion to the
thickness of the catalyst. If the internal stresses in the coating become excessively large, the relatively brittle
Al/Ni intermetallic structure and the fragile nickel residue fail to accommodate for the stresses during heat
treatment and leaching. This will lead to formation of cracks in the developed structure, diminishing the
mechanical strength of the final electrocatalyst. Conversely, by increasing the thickness of the electrocatalyst
a larger actual surface area can be produced. Wendt and Plzak [20] pointed out that only a limited fraction of
the actual surface area of a highly dispersed structure contributes to the electrochemical reaction during
electrolysis. This is because of the limited penetration of the current into the deeper pores of the structure. It
is therefore important to identify the minimum thickness of a porous electrocatalyst without diminishing the
electrocatalytic activity of the final product.
In the present work we investigate how the geometry and structure of a PVD Al/Ni electrocatalyst influences
the actual surface area and the electrocatalytic activity towards the HER and the OER. The structure and
composition of the developed electrodes is characterised by the means of a scanning electron microscope
(SEM) and X-ray diffraction (XRD). The electrocatalytic activity and actual surface area are studied with
potentiodynamic polarisation, cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS).
2 Experimental
2.1 Preparation of electrodes
The electrodes were produced by PVD of aluminium onto a 0.5 mm thick nickel plate (99 wt.%). The nickel
plate was cathodically degreased for two minutes prior to the PVD process. The Al PVD was prepared in a
non-reactive DC-magnetron sputtering mode using a CC800/9 SinOx coating unit from CemeCon AG. In
order to remove nickel oxide from the surface, the nickel was heated and etched in situ by argon sputtering
prior to the sputtering process. The thickness of the aluminium coating was about 20 µm. The aluminium
deposited nickel plates where cut to form three 25×25 mm coupons, suitable for XRD measurements and
electrochemical measurements, and seven 15×10 mm coupons, suitable for microscopy investigations. The
coupons were then heat treated and selectively aluminium leached as outlined in table 1.
Appended Paper II 3
Treatment Ref. Name 25×25 mm coupons 15×10 mm coupons

No treatment AlNi - x
10 min at 610 °C AlNiHT10 - x
10 min at 610 °C + leaching AlNIHT10+L x x
20 min at 610 °C AlNIHT20 - x
30 min at 610 °C AlNIHT30 - x
Table 1: Outline of the heat and leaching treatments for the PVD aluminium deposited nickel coupons.
The heat treatments were prepared in an atmospheric furnace. The selective aluminium leaching procedure
was performed in 30% KOH and 10% KNaC4H4O6*4H2O at 80 °C with stirring for 24 hours.
2.2 Structural characterisation and composition
The un-leached 15×10 mm coupons, listed in table 1, were hot-mounted in PolyFast from Struers, ground
down to 4000 grit and subsequently polished with 3 and 1 µm diamonds. In order not to damage the porous
surface of the leached coupons during mounting, more delicate mounting process was selected. The leached
15x10 mm coupons were cold mounted in epoxy from Struers via vacuum impregnation. The cold mounted
specimens were grinded down to 4000 grit, subsequently polished with 3 and 1 µm diamonds and
mechanical/chemical polished with 0.04 μm colloidal silica.
All the prepared specimens listed in table 1 were investigated in TM 3000 Tabletop scanning electron
microscope from Hitachi. An integrated energy-dispersive X-ray spectroscopy (EDS) was utilised for
elemental analyses. Phase analyses were performed via grazing incidence (GI) XRD on a Bruker AXS, D8-
Discover with Cu Kα radiation. The GI angel was selected to be 6 degrees for all the specimens.
2.3 Electrochemical measurements
The electrocatalytic activity of the developed electrodes was evaluated by means of anodic and cathodic
potentiodynamic polarisation, CV and EIS. The measurements were carried out using a Gamry Reference
3000 potentiostat. A typical three-electrode electrochemical cell made of Teflon and was used for the
electrochemical measurements. The cell contained an Hg/HgO reference electrode from Radiometer
Analytical and a counter electrode of pure nickel. The electrolyte consisted of 1 M KOH and the
measurements were carried out at 25 °C. The potentiodynamic and cyclic voltammetry measurements were
IR-compensated by the means of current interruption. All following half-cell potentials are reported vs.
Hg/HgO.
The experimental procedure for the cathodic polarisation curve was as follows: conditioning at -1.015 V for
30 minutes, open circuit potential (OCP) until 0.01 mV/sec stability followed by a cathodic sweep from -
0.615 V to -1.415 V at a scan rate of 1 mV/s.
The electrocatalytic active surface area of the electrodes was determined by the amount of nickel hydroxide
formed during cyclic voltammetry. For the cathodic reaction, CV measurements were performed from -1.0 V
to -0.2 V at a scan rate of 50 mV/s. In order to eliminate the amount of hydrates from the surface, the
electrodes were conditioned at -1.015 V for 5 minutes prior to the measurements. Following the CV, EIS was
used for determination of double layer capacitance as a measure for the electrochemical active surface area.
4
The measurements were conducted between 100 mHz and 100 kHz at a potential of -1.1 V with an AC
amplitude of 3 mV.
In the case of anodic measurements the procedure was initialised by a 30 min. polarisation at -0.943 V, in
order to facilitate a stable NiO structure. Thereafter, a CV between -0.943 and 0.68 V at the scan rate of 50
mV/s was recorded. Each CV measurement was cycled 10 times in order to obtain reproducibility. The CV
measurements on the samples was designated CVbefore. The CVbefore was followed by a 5 min conditioning at
0.285 V, being approximately the theoretical OER potential. A steady state measurement was then recorded
from 0.285 V to 0.80 V at a scan rate of 1 mV/s in order to measure the OER activity. In order to look for
structural changes this measurement was followed by a second 5 min conditioning and CV, designated
CVafter using the same parameters as previously.
For comparison, anodic and cathodic potentiodynamic polarisation, CV and EIS investigations were
prepared on a 25×25 mm, 3 µm diamonds polished, nickel plate (99 wt.%).
3 Results and discussions
3.1 Structural characterisation and composition
The as-deposited stage
The aluminium deposit obtained from the PVD process is characterised with columnar crystal structure. The
columnar grains are small in width, < 1µm, close to the substrate and become larger as the deposited layer
grows, further away from the substrate. This is verified with a cross section micrograph of a broken Al PVD
coated silicon wafer deposited simultaneously with the nickel substrate in the present work, see fig. 1.
Fig 1: Cross section micrographs of a physical vapour deposited aluminium on top of a silicon wafer. The silicon wafer is broken
mechanically. The micrograph reveals the columnar structure of the deposit. Courtesy of the Technological Institute of Denmark.
Appended Paper II 5
Thermo-chemical diffusion
Fig. 2 shows cross section back scatter electron (BSE) micrographs of the PVD Al deposited nickel
substrates, as-deposited and after 10, 20 and 30 minutes of heat treatment. The EDS analyses from the cross
section micrographs are listed in table 2.
Fig 2: Cross section BSE micrographs of the PVD Al/Ni coatings before and after 10, 20 and 30 min. heat treatments A) as deposited
[AlNi], B) 10 min. heat treatment [AlNiHT10], C) 20 min. heat treatment [AlNiHT20] and D) 30 min. heat treatment [AlNiHT30].
The numbers refer to the EDS analyses reported in table 2.
10 min. heat treatment 20 min. heat treatment 30 min. heat treatment

(EDS no) AlNi AlNiHT10 AlNiHT10+L AlNiHT20 AlNiHT20+L AlNiHT30 AlNiHT30+L
Phase Al Ni Al Ni O Al Ni Al Ni O Al Ni Al Ni O Al Ni
wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.% wt.%
(1)Ni - 100 - 100 - - 100 - - - - 100 100 - - 100
(2)Al3Ni2 - - 41 59 17 4 79 38 61 12 5 83 37 62 18 4 78
(3)Al3Ni - - 55 45 12 - 88 57 42 9 2 89 59 41 11 3 86
(4)Al 100 - 98 2 - - - 97 3 - - - - - - - -
Table 2: Results from the cross section EDS analysis on the specimens in fig. 2 and fig. 5. All elements from the periodic table are
analysed.
6
Relating the EDS results with the Al-Ni binary alloy phase diagram, the diffusion layer no. 2 is predicted to
mainly consist of the Al3Ni2 phase and layer no. 3 of the Al3Ni phase. The XRD analyses on the heat treated
samples (fig. 3) verify the existence of the Al3Ni and Al3Ni2 phases in the PVD Al-Ni diffusion couples.
Fig. 3: GI X-ray diffractograms for the PVD Al-Ni couples after various times of heat treatments. () Al, ()Ni, () Al2O3 , () Al3Ni
() Al3Ni2. JCPDS card-numbers 4-787, 4-850, 46-1212, 3-1052 and 2-416. The incident angel is 6 degrees for all the tested
specimens.
A part from the two intermetallic phases, pure nickel and aluminium peaks are detected. The intensities of
the aluminium peaks are the largest for the shortest diffusion time, as expected. The Al3Ni and Al3Ni2 signals
get more pronounced in proportion to diffusion time. The only non-overlapping pure nickel peak is to be
found at about 93 degrees 2θ. The highest intensity is for 10 min. of diffusion, where nickel has had the least
time to diffuse into the structure. This is presumably scattering signals from the nickel substrate due to the
high penetration depth of x-rays in pure aluminium, resulting in larger amount of scattering from the
substrate. Hence, no assumptions can be made about the amount elemental nickel in the diffusion layers from
the XRD analyses alone.
When the existence of the Al3Ni2 and Al3Ni phases in the interdiffused layers has been verified, the
following can be identified from the micrographs in fig. 2. After 10 min. of heat treatment about 5 µm Al3Ni2
and 4 µm of the Al3Ni intermetallic phases are formed. In the remaining aluminium, closest to the top
surface, particles with higher Z contrast (larger atomic number) compared to the surrounding aluminium are
observed. This indicates diffusion of nickel containing particles into the PVD aluminium coating. Moreover,
the ragged interface of layer 2 into layer 3, and layer 3 into layer 4, seen fig. 2b and c, indicate that the
diffusion mechanism is mainly controlled by the movement of nickel-rich phases into the aluminium and
aluminium-rich phases. These findings are in contrast with the observations of M. Jansssen and G. Rieck
[24], where only aluminium was found to take part in the diffusion mechanism for Ni-Al couples at
temperatures at about 600 degrees. It is however well known that grain boundaries provide high diffusivity
paths in metals and that diffusion along grain boundaries is in order of magnitude faster than bulk diffusion
[25].
Appended Paper II 7
According to Harrison’s classification of the diffusion kinetics [26], diffusion may be considered to take
place only within the grain boundaries at short diffusion times/and or when the volume diffusion coefficient
is much smaller than the grain boundary diffusion coefficient. Due to the large columnar structure of the
aluminium deposits, one or both of this mechanism do evidentially take place during heat treatments of the
PVD aluminium deposited nickel specimens. One could moreover imagine that as the concentration of the
nickel builds up in the grain boundaries, nickel atoms start to diffuse from the boundary into the aluminium
bulk contributing to the formation of the Al3Ni and later the Al3Ni2 phase. Over time the diffusion
characteristic changes from being grain boundary diffusion controlled into being volume diffusion controlled
and competition between the formation of Al3Ni and Al3Ni2 takes place. From the experiments of Castleman
and Seigle [27] the diffusion coefficient of Al3Ni and Al3Ni2 are found to be 1.8 x 10-11 and 9.1 x 10-10
cm2/sec, respectively. Hence, the Al3Ni2 phase grows faster than the Al3Ni phase and therefore, for structures
with limited aluminium sources, when heat treated for long enough time the Al-Ni diffusion film will consist
solely of the Al3Ni2 phase. This is in agreement with the results shown in fig. 2 where the Al3Ni2 phase
increases with extended time of heat treatment. For 20 minutes of heat treatment about 5 µm of Al3Ni and 11
µm Al3Ni2 is formed. When heat treated for 30 min. no aluminium is left on the top surface and almost all
the diffused layer is transformed into the Al3Ni2 phase, resulting in about 18 µm Al3Ni2 and 2 µm of Ni3Al.
8
Selective leaching of aluminium
Fig. 4 and 5 show surface and cross section BSE micrographs of the heat treated specimen after selectively
alkaline leaching of the aluminium. The EDS analyses on the leached cross sections in fig. 5 are listed in
table 2.
Fig. 4: Surface BSE micrographs of the PVD Al/Ni coatings after 10, 20 and 30 min. heat treatment and leaching. A) AlNiHT10+L,
B) AlNiHT20+L and C) AlNiHT30+L.
Appended Paper II 9
Fig. 5: Cross section BSE micrographs of the PVD Al/Ni coatings after 10, 20 and 30 min. heat treatment and leaching. A)
AlNiHT10+L, B) AlNiHT20+L and C) AlNiHT30+L. The numbers refer to the EDS analyses reported in Table 2.
The surface micrographs indicate that the top structure is highly porous. It is evident, from the surface and
cross section micrographs, that the cracks perpendicular to the surface become larger for extended heat
treatments, the cracks are seen as macro pores in the surface micrographs. As there are no cracks in the
structure after heat treatment, the cracks are obviously formed during leaching. The cracks are presumably
due to shrinkage of the lattice during leaching resulting in tensile stresses in the structure. The internal
stresses in the Al3Ni2 layer evidently increase in proportion to the layer thickness.
At the available magnification, for the cross section images in fig. 5, no pores in the leached Al3Ni2 structure
are visible. However, the finding of oxygen and the reduction of aluminium, from about 40 wt.% aluminium
for the un-leached sample to only 5 wt.% for the leached sample, indicate that the leached Al3Ni2 phase is
porous as well. Due to the larger amount of aluminium in the initial Al3Ni structure, the leached nickel
residue becomes considerably more macro porous, compared to the leached Al3Ni2 structure. It is noteworthy
that the residue aluminium after leaching is significantly less in the present specimens compared to the
previous published study [23], where the PVD Al/Ni electrodes were heat treated for 24 hours. In that study,
15 wt.% aluminium residue was observed in the skeletal nickel structure. The deviation in aluminium content
is, thus, probably due to grain growth of the Al3Ni2 crystallites during longer annealing time. The larger
Al3Ni2 crystal structure is evidently not as prone to alkaline leaching as the present one.
XRD diffractograms from the heat treated and leached samples are shown in fig. 6.
10
Fig. 6: GI X-ray diffractograms for PVD Al/Ni coatings after 10, 20 and 30 min. heat treatment and selective aluminium leaching.
The incident angel is 6 degrees for all the specimens. () Ni, () K. JCPDS card-numbers 4-850 and 40-994.
After the leaching procedure, only nickel peaks and potassium peaks, from the leaching procedure, are
identified in the diffractograms. This confirming that most of the aluminium has been leached from the
intermetallic structures as indicated in the EDS analyses. The intensity of the nickel peaks is the largest for
the thinnest sample, 10 min. heat treatment, indicating some scattering from the nickel substrate.
3.2 Electrochemical Measurements
3.2.1. Cathode
Cyclic Voltammetry
The initial cyclic voltammograms for polished nickel and the heat treated and leached coatings are shown in
fig. 7 and fig. 8.
Appended Papper II 111
Fig. 77: Cyclic voltam

mmogram record
ded on polishedd Ni. The anodiic peak represen
nts the oxidation
on of Ni to α-Nii(OH)2.
Fig. 8: Cyclicc voltammograams recorded on o the PVD All/Ni coatings after a heat treatm
ment for 10, 220 and 30 min
n. and selectivee
aluminium leaaching. The anoodic peak repressents the oxidattion of Ni to α-N
Ni(OH)2.
A characterristic of evaluuating the ro

oughness facttor using CV V is that the electrochemi
e ically active surface areaa
is measuredd, that is the surface
s that can w the electtrolyte. Unlikke the Brunaauer-Emmettt
c exchangge electrons with
–Teller theoory (BET), where
w the phyysical absorpption of gas on the surfacce is measure
red. Accordin ngly, the CVV
method seeems more apppropriate co onsidering thhe applicatioon of the eleectrocatalyticc coating. The
T peak forr
formation oof Ni  α-N Ni(OH)2 is useed for the evvaluation andd area under the curve, ii.e. the charg
ge associatedd
with the forrmation of α--Ni(OH)2, waas integratedd by the mean ns of Gamry Echem Anal alyst softwaree. In order too
calculate thhe roughnesss factor the values for each electrode, shown in table 3, were comp pared to thee
-2
measured chharge of 5144 µC·cm forr a polished nickel electrrode establish hed elsewher
ere [28]. The fact that thee
12
area under tthe curve in fig. 7 does not equal 5114 µC·cm-2 is
i due to small residue rooughness aftter polishingg
and uncertaainties from measurement
m ts.
Charge Rf Tafel slo

ope ηHER @
@ 100
E
Electrode
[C·cm-2] [cm2·cm-2] [mV decc-1] [mA cmm-2]
Polished Ni 574E-6 ~1 150 5088
AllNiHT10+L 492E-3 ~800 95 1711
AllNiHT20+L 955E-3 ~1900 82 1311
AllNiHT30+L 1156E-3 ~2200 76 1233
Table 3: Resullts from the cathhodic CVs and potentiodynammic polarisations recorded on th
he PVD Al/Ni ccoatings compaared to polishedd
Ni. The theoreetical potential of
o HER is estim
mated to be -9433 mV vs. Hg/HgO.
The values state the treend from SE EM images iin fig. 5, thaat a significaant leachablle diffusion zone can bee
obtained froom very shorrt thermal treeatments andd that the forrmed intermeetallic phase s increase raapidly due too
the fast difffusion properrties of the PVD
P coating . What is not clear thoug
gh from the B BSE microgrraphs is how w
the surface area increases according g to longer s oaking time – the fast diiffusion kineetics thus giv
ves an effectt
within onlyy 10 min, but b as seen from AlNiiHT20+L an nd AlNiHT3 30+L longerr diffusion timet furtherr
maximizes tthe surface area.
a
Steady-Statee Potential Dynamic

D Sweeep
Steady-statee voltammeetry was ap pplied for m measuring th he electroch hemical actiivity of thee developedd
electrodes. The potentiial was swep pt over a raange compriising the current densitiies traditionally used inn
industrial allkaline electrrolysers [4]. The observeed trend from
m cyclic voltaammetry is m
manifested in
n fig. 9.
Fig. 9: C
Cathodic potentiiodynamic polaarisations recordded on the PVD
D Al/Ni coatings and polished N
Ni in 1 M KOH
H at 25°C.
All three ellectrodes shoow much im

mproved cataalytic activity y for the HE
ER, as comppared to poliished nickel,,
showing a rreduction in overpotentiaal of 337 mVV for AlNiHT T10+L to 3885 mV for A AlNiHT30+L. In betweenn
lays AlNiHHT20+L closer to the AllNiHT30+L results, as could c pected from Rf values. Hence,
be exp H theree
seems to bee an agreemennt between measured
m surrface area an
nd electrochemical activitty.
From fig. 9 it is notewoorthy that thee decline of tthe Tafel slo ope, associated with the rreaction mecchanisms forr
the HER [3,,29], is significantly alterred when com mparing the three samplees to polishedd nickel, i.e. the intrinsicc
properties oof the coatinggs seem to diiffer from puure nickel. It is evident, however,
h thatt the numberr of availablee
reaction sitees is the preddominant facctor for the inncrease in caatalytic activiity. The Tafeel slopes for the cathodicc
potential sw
weeps as welll as overpoteentials for HE ER are showwn in table 3.
Electrochem
mical Impedaance Spectro
oscopy measuurements
In order to assess the electrocatalyt

e tic propertiess of the elecctrodes, impeedance specttra were anaalysed for alll
the electroddes. Electroochemical im mpedance sppectroscopy conducted at -1.1V ((Vs. Hg/HgO O) revealedd
electrocatalyytic differennces between
n the process ed electrodees. In fig. 10 a Nyquist pllot for polish
hed nickel iss
seen while pplots for NiA AlHT10+L, NiAlHT20+L
N L and NiAlH HT30+L are collected
c in ffig. 11.
Fig. 10: Nyyquist plot from

m impedance speectra recorded on
o polished nickel at a potentiaal of -1.1 V.
14
Fiig. 11: Nyquist plots from impedance spectrum

m recorded on the PVD Al/Ni coatings at a ppotential of -1.1 V.
All three im
mpedance sppectra produ uced arcs whhich could be b satisfactorily fitted too the electric equivalentt
circuit (EEC
C). At higherr frequenciess a slight disttortion seem
ms noticeable, without anyy indicationss of a secondd
arc, though.. Observing and fitting of multiple arrcs at higher frequencies has been maade by other authors [30]]
and is norm
mally associatted with chaaracteristics oof the coatin
ng itself from
m one arc, whhile the second is relatedd
to the HER R (or OER). As for the purpose off this study it is not fou und relevantt to increment to higherr
frequencies.
Since only a single arcc appears in the complexx-impedancee-plot the ch haracteristicss of each eleectrode weree
described bby an EEC based
b on a single time constant mo odel used fo
or fitting parrameters, i.e. the classicc
Randles EE EC seen in fig. 12. Here, RCT is th the charge transfer
t resisstance assocciated with thet workingg
electrode, Re is the (ohhmic) resistaance of the electrolyte and Cdl is the
t double llayer capacittance of thee
electrocatalyyst.
Fig. 12: Randles EE

EC used for fiitting EIS paarameters
The choice and fitting procedure of o EEC is off course critiical in orderr to correctlyy describe th he non-ideall
situation at hand. The single
s time-cconstant moddel is a wideely used emp pirical modell to account for the non--
ideal behaviiour of the capacitive eleements due too different phhysical phennomena such as surface heterogeneity
h y
resulting froom surface roughness, immpurities, disslocations etcc. [31]. The roughness faactor has beeen calculatedd
from a com mparison witth the capaccitance meassured from a plane nick kel surface [332] showing g a value off
-2
25µF·cm . The catalytic parameterss obtained frrom the Rand dles EEC fro om the develooped electroocatalysts aree
summarizedd in table 4.
Re Rct Cdl Rf
E
Electrode
m·cm2]
[Ohm [Ohm·cm2] c -2]
[F cm
Poolished Ni 565
5e-3 220 1.49
90e-4 6
A
AlNiHT10+L 368e-3 362e-3 4.40
09e-2 ~900
478e-3 140-3 9.84

47e-2
A
AlNiHT20+L ~2000
457
7e-3 122e-3 11.3
34e-2
A
AlNiHT30+L ~2300
Taable 4: EIS fittiing parameters from

f Randles E
EEC data record
ded on the PVD
D Al/Ni coatingss and polished Ni.
N
Again, the deviance froom 1 for pollished nickell is ascribed to measurem ment uncertaainties and small residuee
roughness after polishiing. The ro oughness vallues for thee developed electrodes attained fro om the EIS S
measuremennts are evideently in agreeement with thhe results fro
om the CV in
nvestigationss.
3.2.2. Anodee
Cyclic Voltaammetry
Fig. 13 andd fig. 14 shhow a comp parison of thhe cyclic vo

oltammogram
ms for the hheat treated and leachedd
coatings, recorded priorr and after a steady
s state m
measuremennt.
Fig. 13: Cyclicc voltammogram

ms recorded on
n the developedd electrocatalyticc coatings priorr the anodic pottentiodynamic measurements
m
compared withh polished Ni.
16
Fig. 14: Cyclicc voltammogram

ms recorded on
n the developedd electrocatalyticc coatings afterr the anodic steaady state potenttial dynamic
sweeps recordded compared with
w polished Nii.
All three tested electroddes show well defined annd symmetriccal peaks forr the reductioon of oxy-hy ydroxide, β -
NiOOH → β-Ni(OH)2 [33]. [ For all three samplees an individ dual peak waas also obserrved on the scan
s towardss
more positivve potentialss indicating reversibility
r oof the reactio
on.
A pronouncced anodic peak

p was fou
und. This is accepted as the reversib
ble Ni(II) → Ni(III) tran
nsformation,,
generally w
written as [34]:
β-Ni(OH)2 + OH- ↔ β-N

NiO(OH) + H2O + e-
ution which iin the case of

The reactionn is followedd by O2 evolu o alkaline meedia is [35,366];
4OH- → O2+4e- + 2H2O.

O
Fig. 13 showws the beforre CVs for all three sampples compareed with nickkel. When coomparing thee before CVss
with the nonn-porous nicckel, an increease in β-NiO
O(OH) → β-NNi(OH)2 charrge were obsserved. By in ntegration ann
increase of the charge associated with the reducttion was foun
nd to be in th
he followingg order; polished nickel >
10 min > 200 min > 30 min,
m see tablee 5.
Appended Paper II 17
Charge Rf ηOER @ 100

Electrode
[C·cm-2] [cm2·cm-2] [mA cm-2]
Polished Ni 1.3E-3 ~1 412
AlNiHT10+L 643E-3 ~500 338
AlNiHT20+L 983E-3 ~750 347
AlNiHT30+L 1429E-3 ~1100 340
Table 5: Results from the anodic CVbefore and potentiodynamic polarisations recorded on the PVD Al/Ni coatings compared to
polished Ni. The theoretical potential of OER is estimated to be 286 mV vs. Hg/HgO
This was found to be in agreement with visual inspections of the coatings thicknesses as discussed earlier.
The porous electrodes have active surface areas in the order of 500 – 1100 times polished nickel. Other
authors state that peak charge for the Ni(OH)2  NiOOH transition is in the order of 0.7 – 1.0 mC cm-2
which corresponds to oxidation of a monolayer on a smooth nickel surface [37],[38].
The proposed explanation for finding lower roughness factor values by looking at β-NiO(OH) ↔ β-Ni(OH)2
peak charges, compared to what was found via Ni ↔ α-Ni(OH)2, is either poor utilization of the inner
branched structure of the high surface coating or a blockage with hydrated oxides of the in-depth
nanostructure. The formation of hydroxide takes place as a consequence of the initial hydration of NiO upon
un-polarized immersion in the electrolyte [38,39].
From the CVs recorded after the steady state experiment the charge tendency changed and the following
order of anodic charge was observed; Ni > 10 min >30 min >20 min (fig. 14). It is believed that this
inconsistency was caused by partial delamination of the coating which also was found on post-SEM images
as illustrated in fig. 15.
Fig. 15: AlNiHT20+L to the left, AlNiHT10+L to the right, subsequent to the OER characterization. Illustration of separation of
phases.
18
Even after tthe steady staate experimeents all samplles indicated

d a significan
ntly higher chharge associaated with thee
direct oxidaation of β-NiiOOH, comp pared to nonn-porous nick kel. This couuld be causedd by at leastt two things;;
the first beiing that the coating
c is on
nly partially delaminated d and the seccond being tthat the undeerlying NiAll
phase has soome beneficcial electrocaatalytic propeerties compaared to polishhed Nickel. T The delaminnation for thee
10 min sam mple is shownn in fig. 15.
Steady-Statee Potential Dynamic

D Sweeep
The anodic catalytic acttivity measu urements are shown in fig. 16 where the steady sstate characterization forr
the electroddes is presentted in compaarison with p olished nick
kel. The orderr of activity ccorresponds to what wass
observed byy the evaluatiion of the ch
harge associaated with the anodic peakk from the aft fter-CVs.
F
Fig. 16: Anodicc potentiodynam t PVD Al/Ni coatings in 1 M KOH at 25 C
mic polarisationns recorded on the C.
The overpootential for thhese coating gs was foundd to be 340 mV, at 100 mA/cm2, coompared to 412 mV forr
polished nicckel. The ordder of magniitude for the found overp potentials does seem reassonable low compared too
what is founnd in literatuure. A list off overpotentiaals for well-k
known OER catalysts canan be found in
i i.e. [3,40]..
The recordeed overpotenntials are listeed in Table 55.
Corrosion m
mechanism during
d the OE
ER
With basis iin the observved delaminaation a mechhanism invollving the inittial corrosionn of the coatiing interfacee
is proposedd. We believve that the phase
p transiition interfacce is a weak
k skeletal zoone which upon
u anodicc
polarisationn corrodes beecause of onee or both of tthe followingg corrosion mechanisms:
m
a) Initial disssolution of nickel

n due to
o low Ni-ion concentratio
ons in the eleectrolyte.
b) Localisedd acid formaation inside th

he porous strructure durin
ng oxygen ev
volution.
The former mechanism (a) the corrosion could take place as a follows. In
nitially, wheen the anode is placed inn
the electrollysis cell noo nickel ionss are presennt in the eleectrolyte. Un
nder these cconditions, nickel
n is nott
thermodynaamically stabble at zero po
otential and dduring the OEER and the electrode
e corr
rrodes accord
ding to:
Ni + 3OH-  Ni(OH)3- + 2e-
gram in fig. 117 for nickell ionic concentration at 1ee-6 m.

This can be seen from thhe E-pH diag
Fig 17: E-pH diagram showiing thermodynaamics of Ni in an aqueous sollution, calculateed for 1 atm., 225C with Ni molality
m of 1e-6.
-2
and of 1e .
However, thhe kinetics of

o the nickel dissolution iis most likely
y not fast clo
ose to zero ppotential. Nev
vertheless ass
the current is raised too more positive values,, current is applied bettween the el electrodes, th
he corrosionn
mechanism gets accelerrated by the OER. Thus, the concenttration of nicckel ions witthin the boun ndary regionn
increases annd corrosionn is slowed down and eeventually sttopped. This is evident bby looking at the E-pH H
-2 -
diagram forr nickel ionicc consecratio
on at 1e m wwhere Ni(OH H) 3 cannot bee formed. Heence, only NiiO is presentt
in the regionn between Ni
N and NiOOH H.
For the latteer mentionedd corrosion mechanism

m ((b) it is supp
posed that when protons are formed during OER
R
according too:
2H2O  O2 + 4H+ + 4e-
This reactioon, which corresponds to

t the OER reaction in n acidic wateer electrolyssis, is a con nsequence off
-
disturbance from oxygeen bubbles on
n OH migraation into thee inner activee sites of thee catalyst. Ass a result thee
20
pH is locally decreased. As this happens the skeletal zone is further weakened and corrosion will ultimately
lead to delamination.
The proposed result is that the electrolyte in the pores becomes locally acidic and the Al3Ni structure
corrodes according to:
Al3Ni + 11H+ + 2.75O2(g) = 3Al+3 + Ni+2 + 5.5H2O ∆G0 = -2641 kJ
The gips free energy for the reaction above is highly negative which means that the corrosion mechanism is
highly thermodynamically favourable.
Due to the small nickel versus volume ration for the leached Al3Ni structure, it is particularly vulnerable to
corrosion. That is, only small amount of nickel dissolution results in dramatic weakening of the structure.
The delamination of the leached Al3Ni structure is proposed to be further accelerated by the subsequent
oxygen evolution which causes a coating blow-off as shown in fig 15.
4 Conclusions
Microscopic investigations indicate that nickel and nickel-rich phases are the main moving species during
interdiffusion of PVD deposited aluminium and nickel substrate at 610 °C and not vice versa as might be
expected. This diffusion mechanism leads to formation of fine grained Al/Ni intermetallic structures that are
very susceptible to alkaline leaching of aluminium in strong alkali. By altering the diffusion time, from 10 to
30 min, different thicknesses of intermetallic structures can be tailored. By alkaline aluminium leaching of
the intermetallic structure, a porous nickel electrode, with max 5% aluminium residue, is produced.
The roughness factor of the porous nickel coatings rises in proportion to the thickness of the intermetallic
layer formed, which in addition is in line with the diffusion time. In that way one can obtain high
electrocatalytic surface areas with a roughness factor close to 1000, using the shortest diffusion time, i.e. 10
min. Increasing the diffusion time to 20 minutes will again dramatically increase the roughness factor and
lower the charge transfer resistance associated with the electrode. As the diffusion time is further prolonged
(30 min.) still the electrode characteristics are improved, although to a minor degree. All in all, time and
energy consumption can be lowered in the processing of the electrodes leading to the feasibility of industrial
implementation. What is more, since the material characteristics of the three binary alloys are the same, these
observations clearly emphasize how the electrochemically active surface area plays a dominant role when
designing electro catalytic electrodes.
The leached Al3Ni structure was not stable and became delaminated during the OER. Two corrosion
mechanism accelerated by the O2 gas erosion where presented for the observed delamination. A more
detailed study of this delamination is most certainly needed if these coatings are to be used as anodes for
bipolar electrodes in industrial alkaline water electrolysis. From the OER measurement, it was found that all
coatings comprise higher activity compared to polished nickel. This finding is a reality, even though the
coatings are highly damaged by the polarization, and suggests that the underlying phase provides some
increased activity for the OER. A more detailed study of this underlying phase should therefore be initiated.
Appended Paper II 21
Acknowledgement
The authors wish to thank Melanie Röefzaad at Siemens A/S for assisting with the electrochemical
measurements. The authors also want to thank Lars Pleth Nielsen and Kristian Rechendorff at The Danish
Technological Institute for assisting with the PVD coatings. Ewa Adamsen, Lars Pedersen, John C. Troelsen
and Steffen S. Munch at DTU are all gratefully acknowledged for their invaluable help. Financial support
from The Energy Technology Development and Demonstration Program in Denmark (EUDP) (project
number: 63011-0200) is also gratefully acknowledged.
References
[1] Corrosion resistance of nickel and nickel-containing alloys in caustic soda and other alkalies. INCO
(Int Nickel Co, Inc) Corros Man 1975.
[2] Rebak RB. Nickel Alloys for Corrosive Environments. Advanced Materials and Processes
2000;157:37–42.
[3] Zeng K, Zhang D. Recent progress in alkaline water electrolysis for hydrogen production and
applications. Progress in Energy and Combustion Science 2010;36:307–26.
[4] Ursua A, Gandia LM, Sanchis P. Hydrogen Production From Water Electrolysis: Current Status and
Future Trends. Proceedings of the IEEE 2012;100:410 –426.
[5] Hoogers G. Fuel Cell Technology Handbook, United States of America: CRC Press; 2002, p. 20–5.
[6] Balej J, Divisek J, Schmitz H, Mergel J. Preparation and properties of raney-nickel electrodes on Ni-Zn
base for H2 and O2 evolution from alkaline-solutions. Part I: electrodeposition of Ni-Zn alloys from
chloride solutions. Journal of Applied Electrochemistry 1992;22:705–10.
[7] Oden LL, Russell JH, Sanker PE. Method for producing supported Raney nickel catalyst. United States
patent US4049580. 1976
[8] Raney M. Method of producing finely-divided nickel. . United States patent US1628190. 1927.
[9] Bagotsky VS. Fuel Cells: Problems and Solutions, John Wiley & Sons; 2012, p. 209–10.
[10] Sillitto SMA, Adkins NJE, Ormerod RM, Paul E, Hodgson DR. Characterisation of advanced Raney
nickel electrocatalytic coatings produced by the direct spraying method. Rugby: Inst Chemical
Engineers; 1999.
[11] Sillitto SMA, Adkins NJE, Hodgson DR, Paul E, Ormerod RM. Electrochemical testing and structural
characterisation of nickel based catalytic coatings produced by direct spraying. In: Lednor PW,
Nagaki DA, Thompson LT, editors. Advanced Catalytic Materials-1998, vol. 549, Warrendale:
Materials Research Society; 1999, p. 23–9.
[12] Kellenberger A, Vaszilcsin N, Brandl W, Duteanu N. Kinetics of hydrogen evolution reaction on
skeleton nickel and nickel-titanium electrodes obtained by thermal arc spraying technique. Int J
Hydrog Energy 2007;32:3258–65.
[13] Kellenberger A, Vaszilcsin N, Brandl W. Roughness factor evaluation of thermal arc sprayed skeleton
nickel electrodes. J Solid State Electrochem 2007;11:84–9.
[14] Birry L, Lasia A. Studies of the hydrogen evolution reaction on Raney nickel-molybdenum
electrodes. J Appl Electrochem 2004;34:735–49.
[15] Kellenberger A, Vaszilcsin N. The determination of the roughness factor of skeleton nickel electrodes
by cyclic voltammetry. Rev Chim 2005;56:712–5.
[16] Schiller G, Henne R, Borck V. Vacuum plasma spraying of high-performance electrodes for alkaline
water electrolysis. JTST 1995;4:185–94.
[17] Fournier J, Miousse D, Legoux JG. Wire-arc sprayed nickel based coating for hydrogen evolution
reaction in alkaline solutions. Int J Hydrog Energy 1999;24:519–28.
[18] Boruciński T, Rausch S, Wendt H. Raney nickel activated H2-cathodes Part II: Correlation of
morphology and effective catalytic activity of Raney-nickel coated cathodes. J Appl Electrochem
1992;22:1031–8.
22
[19] Wendt H, Hofmann H, Plzak V. Anode and cathode-activation, diaphragm-construction and

electrolyzer configuration in advanced alkaline water electrolysis. Int J Hydrog Energy 1984;9:297–
302.
[20] Wendt H, Plzak V. Electrocatalytic and thermal activation of anodic oxygen- and cathodic hydrogen-
evolution in alkaline water electrolysis. Electrochimica Acta 1983;28:27–34.
[21] Schiller G. High performance electrodes for an advanced intermittently operated 10-kW alkaline water
electrolyzer. Int J Hydrog Energy 1998;23:761–5.
[22] Maunowski P, Jtilich C. Improved components for advanced alkaline water electrolysis. Int J Hydrog
Energy 1988;13:141–50.
[23] Kjartansdóttir CK, Nielsen LP, Møller P. Development of durable and efficient electrodes for large-
scale alkaline water electrolysis. Int J Hydrog Energy 2013;38:8221–31.
[24] Jansssen M, Rieck G. Reaction diffusion and kirkendal-effect in nickel aluminium system.
Transactions of the Metallurgical Society of AIME 1967;239:1372–85.
[25] Inderjeet Kaur, Mishin Y, Gust W. Fundamentals of grain and interphase boundary diffusion.
Chichester; New York: John Wiley; 1995.
[26] Harrison LG. Influence of dislocations on diffusion kinetics in solids with particular reference to the
alkali halides. Trans Faraday Soc 1961;57:1191–9.
[27] Castleman LS, Seigle LL. Layer growth during interdiffusion in aluminum-nickel alloy system. Met
Soc AIME -- Trans 1958;212:589–96.
[28] Machado SAS, Avaca LA. The hydrogen evolution reaction on nickel surfaces stabilized by H-
absorption. Electrochimica Acta 1994;39:1385–91.
[29] Guerrini E, Trasatti S. Recent developments in understanding factors of electrocatalysis. Russ J
Electrochem 2006;42:1017–25.
[30] Herraiz-Cardona I, Ortega E, Pérez-Herranz V. Impedance study of hydrogen evolution on Ni/Zn and
Ni–Co/Zn stainless steel based electrodeposits. Electrochimica Acta 2011;56:1308–15.
[31] Marceta Kaninski MP, Miulovic SM, Tasic GS, Maksic AD, Nikolic VM. A study on the Co–W
activated Ni electrodes for the hydrogen production from alkaline water electrolysis – Energy saving.
Int J Hydrog Energy 2011;36:5227–35.
[32] Šimpraga RP, Conway BE. The real-area scaling factor in electrocatalysis and in charge storage by
supercapacitors. Electrochimica Acta 1998;43:3045–58.
[33] Kumar M, Awasthi R, Sinha a. SK, Singh RN. New ternary Fe, Co, and Mo mixed oxide
electrocatalysts for oxygen evolution. Int J Hydrog Energy 2011;36:8831–8.
[34] Weininger JL, Breiter MW. Effect of Crystal Structure on the Anodic Oxidation of Nickel. Journal of
The Electrochemical Society 1963;110:484.
[35] Sanchis P, Ieee M. Water Electrolysis : Current Status and Future Trends. Proceedings of the IEEE
2012;100.
[36] Pletcher D, Li X. Prospects for alkaline zero gap water electrolysers for hydrogen production. Int J
Hydrog Energy 2011;36:15089–104.
[37] Nelson PA, Elliott JM, Attard GS, Owen JR. Mesoporous Nickel / Nickel Oxidesa Nanoarchitectured
Electrode. Chem Mater 2002:524–9.
[38] Li X, Walsh FC, Pletcher D. Nickel based electrocatalysts for oxygen evolution in high current density,
alkaline water electrolysers. Physical Chemistry Chemical Physics 2011;13:1162.
[39] Rebouillat S. Paving the Way to The Integration of Smart Nanostructures: Part II: Nanostructured
Microdispersed Hydrated Metal Oxides for Electrochemical Energy Conversion and Storage
Applications. International Journal of Electrochemical Science 2011;6:5830–917.
[40] Hamdani M. Co3O4 and Co- Based Spinel Oxides Bifunctional Oxygen Electrodes. International
Journal of Electrochemical Science 2010;5:556–77.
Appended Paper III 1
Investigations of the diffusion mechanism of PVD Al and Ni couples at

610˚C
Cecilía Kjartansdóttira, Hossein Alimadadib, Takeshi Kasamab and Per Møllera.
a Department of Mechanical Engineering, The Technical University of Denmark
b
Center for Electron Nanoscopy, The Technical University of Denmark
Abstract
PVD Al-Ni diffusion couples are heat treated at 610C for few minutes up to 24 hours. Initially, the
dominant diffusion mechanism is found to be grain boundary diffusion of Ni-rich phases into the high
containing Al structure. It is proposed that the first intermetallic phase to form is AlNi3, appearing as small
crystallites in the grain boundaries of the columnar aluminium structure. Together with the AlNi3 particles in
the aluminium residue, only Al3Ni and Al3Ni2 phases are formed in the diffusion zone for up to 30 minutes
of heat treatment. 2 hours of heat treatment results in depletion of the Al and the Al3Ni and thin layers of
AlNi and AlNi3 are formed closest to the Ni substrate. Also, highly porous -Al2O3 has formed on the top
surface of the Al2Ni3 phase. No remaining AlNi3 are found in the Al3Ni2 structure after 2 hours of diffusion.
Longer annealing time, results in slow enlargement of the AlNi, AlNi3 and -Al2O3 layers and grain growth
in the Al2Ni3 phase.
1 Introduction
Increasing the surface area and altering the electrocatalytic configuration of an electrocatalyst by selectively
leaching one or more element from metal alloys, has widely been used to promote the activity of hydrogen
electrocatalysts [1–11]. Various techniques for producing the leachable NiZn or NiAl alloys, such as
electrodepositiont, powder pressing and thermal spraying have been proposed [3],[9],[7], [12–22].
In a previous study [1], efficient hydrogen electrodes were produced by physical vapour deposition (PVD) of
Al onto a nickel substrate, followed by heat treatment for 24 hours at 610 C and selective aluminium
leaching. The final electrocatalytic behaviour of the electrode depends on the electrode composition and
structure. In order to facilitate the optimal electrocatalysts produced in this, manner understanding the
diffusion mechanism of the PVD Al-Ni couples is essential.
According to the Ni-Al binary alloy phase diagram [23] four thermodynamically stable alloys, namely Al3Ni,
Al3Ni2, NiAl, and Al Ni3, can be formed at temperatures below 854 C. Formation of the Al3Ni5 phase is also
proposed at temperatures below 700 C. The existence of that phase is, however, still not well established.
Several work on Ni-Al couples at temperatures around 600 C have been published [24-27] where only
Al3Ni and Al3Ni2 phases were found to be formed. In one publication [28] thin layers of NiAl and AlNi3
were formed after 340 hours of heat treatment. Janssen and Rieck [24] reported that only aluminium was
found to take part in the diffusion mechanism for Ni-Al couples at temperatures at about 600C. Here, AlNi3
is proposed to be the initial phase to form at 610 C and thin layers of NiAl and AlNi3 are detected after only
2 hours of diffusion. Furthermore, Ni and Ni-rich phases appear to be the main moving species at 610 C in
PVD Al-Ni couples.
2
In this paper we report studies on PVD Al-Ni diffusion couples, heat treated at 610 C for few minutes up to
24 hours, investigated by the means of, high resolution scanning electron microscope (HR-SEM), energy
dispersive X-ray spectroscopy (EDS), X-ray diffractometry (XRD), electron backscatter diffraction (EBSD),
ion channelling contrast imaging (ICCI) and transmission electron microscopy (TEM).
2 Experimental
2.1 Material
Commercially available, rolled and annealed, nickel plate is served as substrate for aluminium physical
vapour deposition (PVD). Using optical emission spectroscopy, the purity of the plate is measured to be
99%. An approximately 20 µm thick layer of aluminium is physical vapour deposited onto the Ni substrate
by the means of non-reactive DC-magnetron sputtering process, using a CC800/9 SinOx coating unit from
CemeCon AG. The aluminium source is an Al 1050 alloy target run at 750 W, the RF bias on the substrate is
set to 800 W and the start pressure in the chamber is 1 mPa. Before placing the nickel plate inside the PVD
chamber the plate is cathodically decreased and dried. Nickel oxides are removed from the surface in situ by
means of Ar sputtering. The Al deposited Ni plate is cut to form 25×25 mm2 coupons, suitable for X-ray
diffraction measurements and 10×15 mm2 and 5×10 mm2 coupons suitable for microscopy investigations.
The prepared coupons are heat treated in an atmospheric furnace at 610˚C for various times: 0, 3, 10, 20, 30,
120, 240, 360, 480, and 1440 min. The coupons are removed from the furnace directly after heat treatment
and cooled at ambient temperature.
2.2 Sample preparation
For cross section investigations the 10×15mm2 coupons are hot-mounted in PolyFast from Struers grinded
down to 4000 grit and subsequently polished with 3 and 1 µm diamond. For detailed microscopic
characterisation the 5×10 mm2 coupons are mounted in a custom made sample holder specially prepared for
cross section polishing. The coupons are grinded down to 4000 grid followed by 3, 1 and 0.25 µm diamond
polishing and mechanical/chemical polishing with 0.04 μm colloidal silica (OPS from Struers).
2.3 Grazing incidence X-ray diffraction
The grazing incident X-ray diffraction (GI-XRD) method is applied for phase analyses. The analyses are
performed using a Bruker axs, D8-Discover instrument with Cu Kα radiation. The GI angel is selected to be
6 degrees for all the XRD investigations.
2.4 Electron / Ion microscopy
For general microstructural and elemental investigations a Hitachi TM 3000 Tabletop SEM is applied for all
samples. For detailed investigations of the as-deposited sample and samples heat treated for 10 and 120
minutes, electron backscatter diffraction (EBSD) and ion channelling contrast imaging (ICCI) are performed
in a FEI Helios NanoLabTM 600, equipped with an EDAX-TSL EBSD system and a Hikari camera. The ICCI
investigation is performed using Ga+ ions with energy of 30 keV.
The sample heat-treated for 120 minutes is studied in an FEI Titan 80–300 filed-emission-gun TEM,
equipped with an energy dispersive X-ray spectroscopy (Oxford Instruments, X-MaxN 80 mm2 silicon drift
detector) and a spherical aberration probe corrector. The microscope is operated in scanning TEM mode at
300 kV and the images are acquired using either a bright-field or a high-angle annular dark-field detector.
STEM-EDS is carried out with an interval of 10 nm from a region of 100×200 nm2.
The EBSD measurement is performed in a hexagonal grid with an electron probe current of 5.5 nA at an
acceleration voltage of 15 kV. The cleaning procedure of the measured data is applied using OIM 6TM as
follows: (i) grain confidence index standardization, (ii) single iteration grain dilation (in both cases, a grain is
defined as a region consisting of at least four connected points with misorientations of less than 5˚), (iii) all
the data points with confidence index below 0.1 are disregarded.
3 Results
3.1 General investigations
Backscatter electron imaging
Fig.1 shows backscatter electron (BSE) micrographs of the cross section of PVD aluminium nickel coupons
as- deposited and after various times of heat treatments at 610˚C. The micrographs clearly show the general
trend of microstructural change as a function of time. In the as-deposited state the large difference in the BSE
intensity between aluminium and nickel is evident. This owes to higher atomic number of nickel which
appears brighter compared to aluminium with lower atomic number (Z contrast). Z contrast is beneficial for
general investigation of the change in microstructure upon heat treatment.
As can be easily seen in Fig.1, after only 3 minutes of heat treatment, particles with higher atomic number
than Al appear within the Al layer. For the coupon heat treated for 10 minutes, a larger amount of bright
particles are detected in the Al top layer. Furthermore, two sharply defined layers, referred hereafter as W
and Z, are formed between the nickel substrate and the Al layer. With further increase of the diffusion time,
an increase in the formation of Z layer on the detriment of the W phase is observed. After 30 minutes of the
heat treatment, no pure Al layer is left and only the diffusion zone and the intermetallic structure, consisting
primarily of the Z phase, remains.
4
Al
Bright Al
particles
W
Al Al W
Z Z
10 µm Ni 10 µm Ni 10 µm
10 µm
Ni
As deposited 3 min. 10 min. 20 min.
Porous layer Z
W
Z
Y Y Z
Z X
X
10 µm Ni 10 µm Ni 10 µm Ni 10 µm
30 min. 120 min. 240 min. 1440 min.
Fig.1: Backscatter electron images on the cross section of the PVD Al/Ni coupons after various times of heat treatment at 610˚C.
Heat treatment for 120 minutes brings about formation of a porous layer at the very top surface and two
newly formed layers close to the nickel substrate, hereafter entitled X and Y. It is noted that BSE
micrographs vaguely reveals presence of equi-axed grains in the Z layer.
Heat treatment for longer times, i.e. 360, 480 and 1440 results in a slow growth of X and Y layers as well as
growth of equi-axed grains in the Z layer. It is noted that micrographs for the 360 and 480 minutes coupons
do not contribute with any new microstructural information. Investigations of these are therefore considered
superfluous and results thereof will not be included here. In order to investigate the chemical composition in
the diffusion layers of the Al-Ni couples, EDS and GI-XRD are applied.
Energy dispersive X-ray spectroscopy
To characterize different diffusion layers revealed by backscatter electron imaging, EDS is applied on the
cross-section of the samples shown in Fig. 1. The chemical compositions of Ni in the detected diffusion
layers are listed in Table 1.
Time of
W [Al3Ni] Z [Al3Ni2] Y [AlNi] X [AlNi3]
diffusion
(wt.% Ni) (wt.% Ni) (wt.% Ni) (wt.% Ni)
(Min.)
10 45 59 - -
20 42 61 - -
30 41 62 - -
120 - 63 x x
240 - 64 71 87
1440 - 64 71 88
Table 1: The chemical composition of Ni in the PVD Al-Ni diffusion layers shown in Fig 1. The possible binary alloy is indicated in
square brackets.(– The phase is not present. x The interaction volume of electron beam is larger than the width of the layer.)
It is noted that the chemical composition of all the cases is balanced to 100 by Al. Relating the EDS results
with the Al-Ni binary alloy phase diagram the diffusion layers are predicted to mainly consist of the
following intermetallic phases: W, Z, Y and X are Al3Ni, Al3Ni2, AlNi and AlNi3 respectively.
Grazing incidence X-ray diffraction
In order to supplement the EDS information, GI-XRD is applied. The results from the XRD measurements
are shown in Fig. 2.
6
Fig. 2: GI X-ray diffractograms for the PVD Al-Ni couples after various times of heat treatments. () Al,  Ni, () Al2O3
(Corundum), () Al3Ni () Al3Ni2. JCPDS card-numbers 4-787, 4-850, 46-1212, 3-1052 and 2-416. The incident angel is 6 degrees
for all the tested specimens.
From the diffraction patterns for the as plated and 3 min. coupons aluminium, aluminium oxide and nickel
diffraction peaks are detected. On grounds of that the as plated PVD Al structure does not contain any Ni, the
Ni signals in the XRD result must stem from the Ni substrate. Therefore, although only 6 degrees GI angle is
applied, X-ray diffraction from the Ni substrate can be expected due to the large penetration depth of X-ray
in Al. Consequently, no interpretations can be made about the presence of pure Ni in the diffusion layer of
the heat treated coupon from the XRD results alone. While it is evident, from the SEM image, that some
diffusion has taken place into the PVD Al layer after only 3 minutes of heat treatment, no intermetallic phase
is detected from the corresponding diffractogram.
For the coupon heat treated for 10 min. Al3Ni and weak Al3Ni2 diffraction peaks are detected in addition to
those previously found. For the 20 and 30 min. coupons the Al3Ni and Al3Ni2 peaks get more pronounced.
Also, the Ni diffractions at ~76˚ 2θ and ~98˚ 2θ disappear. This is most likely due to a smaller amount of X-
ray signals from the Ni substrate because of less X-ray penetration in the intermetallic phases formed,
compared to pure Al. After 120, 240 and 1440 minutes of heat treatment, higher intensities from the Al3Ni2
phase are detected and all the Al3Ni and Al2O3 peaks disappear.
As seen in Fig. 2, a number of low intensity diffractions, for the coupons heat treated for 10 min. and longer,
have not been assigned to any phases. Due to many overlapping peaks, distinguishing between different Al-
Ni binary alloys and oxides thereof from the XRD results is challenging. When making phase analyses on
the data some of the reaming peaks could correspond to Al4Ni3, NiAl2O4, Al2O3, AlNi and AlNi3. Another
limitation of the XRD analyses is that small crystallites have less diffraction volume compared to larger
ones; therefore, the corresponding peaks have low intensities. Also, small crystals results in a broad
diffraction peak whereas large crystals give sharp diffraction peaks. Therefore, when identifying small
crystallites among diffractions from larger crystallites some diffraction peaks can be overlooked.
Accordingly local microscopic characterisation is carried out on selected samples.
3.2 Detailed investigations
Based on the backscatter electron imaging results from the EDS and XRD investigations, three samples are
chosen for detailed microscopic characterization (as-deposited, 10 min. and 120 min. heat treated). As-
deposited microstructure is selected to identify the crystal structure of the PVD Al before heat treatment. The
10 min. treated coupon is selected to analyse the initial diffusion behaviour and the two firstly detected
diffusion layers i.e. W and Z. Lastly the 120 min. treated coupon is selected to investigate the changes of the
crystal structures in the Z layer together with the latter developed diffusion layers, Y, X and the porous one.
These three samples encompass all the different layers and phases which are formed in the as-deposited state
and after heat treatment for various times.
As-deposited PVD Al
Fig.3 shows the microstructure of the as-deposited Al on Ni.
Al
Ni
Fig. 3: ICCI micrograph of an as deposited PVD Al on Ni, showing the columnar structure of the aluminium.
The Al layer has a columnar microstructure in which the microstructure consists of fine grains at the vicinity
of the substrate and by increase of distance from the substrate, some grains outgrow the rest. Notably, the
microstructure of Al is not fully dense and no epitaxial relation between Ni substrate and Al deposit is
observed.
8
10 min. heat treatment
The cross section forward scatter detector (FSD) micrograph of the PVD Al-Ni diffusion couples after 10
min. of heat treatment is shown in Fig. 4. The micrograph reveals the especially small crystallites in the W
and X layers, compared to the Al and Ni structures.
Al
W
Z
Ni
Fig.4: FSD micrograph of the PVD Al-Ni diffusion couples after 10 min. of heat treatment at 610 C showing the small crystallites in
the W and X layers and the high z-contrast particles in the grain boundaries of the Al.
Image capture from ICCI investigations on the microstructure of the 10 min. sample is shown in Fig. 5. No
additional information could be obtained on Ni and Z layer from the imaging, hence, only the W and the Al
layer are shown in Fig 5. Very fine crystallites are present in the W layer and are marked by arrows in Fig. 5.
Observing the interface between the W layer and the Al, the W layer has an outward curvature whilst the Al
has inward curvature. This strongly suggests that diffusion is from the W layer into the Al layer. In addition,
a nickel rich phase is formed on the grain boundaries and triple lines of the Al columnar grains.
Fig. 5: ICCI micrograph of PVD Al-Ni diffusion couples after 10 min. of heat treatment at 610C.
To characterize the nickel rich phase at the grain boundaries EBSD is applied. Fig. 6a shows the electron
backscatter pattern (EBSP) of one of the nickel rich crystallites detected and Fig. 6b shows the same pattern
after background subtraction. The high quality pattern shown in Fig. 6b is used for phase identification in
which various phases i.e. Al, Al3Ni, Al3Ni2, Al4Ni3, AlNi, Al3Ni5, AlNi3, and Ni are indexed. Indexing the
pattern using AlNi3 provides the highest similarity to the measured pattern as shown in Fig.6c. It is also
attempted to identify the fine crystallites present in the W layer. However, due to the limited diffracting
volume, it is not possible to obtain high quality patterns that can be used for reliable phase identification.

Fig. 6: a) EBSP from one nickel rich crystallite detected in Fig. 4. b) The EBSP after background subtraction. c) Indexing of the
pattern with AlNi3.
Simultaneously to the EBSD, EDS and orientation microscopy are applied. The results are shown in Fig 7.
10

a) EDS map of the diffsu
uion sone. b) EBSD orientation map show
wing the colum
mnar Al
– Ni, –ZZ, –W, – Al structu
ure.

c) Misorientattion angle of the Al depossit.

Fig. 7: EDS
S, EBSD and orrientation microoscopy of the PVD
P Al-Ni diffu
usion couples aafter 10 min.
Fig. 7a show ws the EDS map

m of nickeel covering N Ni, Z, W andd Al layers an
nd Fig. 7b shhows the orieentation mapp
of the sampple at the sam
me location. The
T Ni, Z, W and Al lay yer are clearly
y visible in tthe EDS mapp and appearr
as red, yelloow, green annd blue respeectively in thhe used colou
ur coding schheme. No relliable patternn is obtainedd
in layers Z and W (verry fine crysttallites or am morphous) fo or phase identification, cconsequentlyy orientationn
microscopyy (Fig.7b) yiields no ressults. Howevver, the Al microstructu ure is clearlly visible in
n which thee
columnar nnature of as-ddeposited Al is preserveed. Most of the grains are a close to bblue colour indicating a
rather weakk <111> textuure in the Al deposition ggrowth directtion. The misorientation angle distrib bution is alsoo
calculated ((Fig.7c). Thee distribution
n is close to rrandom distribution of FC
CC material [29], howev ver, there is a
significantlyy higher poopulation thaan random close to 60 0˚ which co orresponds to misorientation of Ʃ33
boundaries in FCC mateerial.
120 min. heeat treatmen

nt
Using ICCII, Ni substratte and the diiffusion layeers X, Y and Z can clearlly be seen inn the sample heat treatedd
for 120 minnutes, see Figg. 8a.

Z
Y
X
Y (not clear)
Ni
X
Ni

a) EBSD image quality mapping. b) ICCI image
Figure 8: Images of PVD Al-Ni diffusion couples after 120 min.
The mean grain size in the X layer is about 300 nm. The microstructure of Y and Z layer are, however, not
visible in the ICCI micrograph. Fig. 8b shows the image quality map of EBSD data in which the
microstructure of Y and Z layers can be easily seen. The average grain size in layer Z is 812 nm. There is a
very limited number of grains in the Y layer in the analysed map, nevertheless the grain size is of the same
order as the grains in layer Z. It is noted that the grains at the top of Z layer, are elongated whilst the rest of
microstructure of the Z layer is composed of equi-axed grains. EBSD (Fig. 8a) and TEM (not shown)
analyses of elongated grains and equi-axed grains in Z layer show that both of their grains are trigonal
Al3Ni2.
A high-resolution STEM image shown in Fig. 9, which corresponds to Al3Ni2 [211], indicates that there are
generally no detectable defects or AlNi3 crystallites in the Z layer (Fig.9a) although several dislocations in a
few grains are observed as shown in Fig. 9a.
12

a) b)
Fig. 9: TEM micrographs of the Z layer for PVD Al-Ni diffusion couples heat treated for 120 minutes. (a) Low-magnification STEM
BF image. Most of the grains do not contain defects or inclusions. (b) High-resolution STEM BF image. The contrast variation
originates mainly from milling damage during FIB preparation.
The porous layer on the top of the sample after 120 min treatment is also investigated with TEM. As shown
in Fig.10a, two different types of the pores are observed. Smaller pores (i.e. voids) have sharp planar
interfaces, while larger pores (>50 nm) have irregular shapes with a rim of ~15 nm showing darker contrast.
STEM-EDS measurements of a pore of about 100 nm in diameter (Fig. 10b) shows that the rim of the pore is
rich in Al and O and is lacking in Ni. A high-resolution TEM image of the rim shown in Fig. 11 suggests the
phase to be gamma-Al2O3.
b)
a)

Fig. 10: a) STEM HAADF image of a porous layer located on the top of the specimen. b) STEM HAADF image and STEM-EDS
elemental maps of a rim of a pore with ~100 nm in size. The elemental mapping shows a composition of 38 at% Al and 62 at% O
(assumed the density to be 4.0 g/cm3 and the thickness to be 50 nm).

a)
b) c)
Fig.11: a) High resolution STEM BF image of a rim of a pore of ~100 nm in size. The lattice spacing of 0.29 nm at the bottom-left
corner corresponds to {101} planes of Al3Ni2. Pt particles on the surface of the rim appear to be transferred from a Pt protection layer
by FIB milling. Wide lattice fringes observed at the rim are in good agreement with those of gamma-Al2O3 [011]. (b) Fast Fourier
transform of an area marked with a box in Fig11a, (c) Simulated electron diffraction pattern of gamma-Al2O3 [011].
4 Discussion
As evident from the ICCI micrograph in Fig. 3, the as-received PVD aluminium coating is characterised by a
not fully dense, void open boundaries, columnar microstructure. This type of coating are typically formed
with vapour deposition where the substrate temperature is relatively low compared to the melting
temperature of the coating [30],[31] and may indicate tensile intrinsic stresses in the coating [32].
From the BSE cross section micrographs investigations shown in Fig. 1 nickel rich particles are observed to
diffuse in to the PVD aluminium coating after only 3 minutes of heat treatment. EBSD investigations on the
10 minutes heat treatment sample, Fig. 5, indicate the nickel rich phase to be AlNi3 (Fig. 6). The formation of
high nickel containing phase instead of a high Al containing phase inside the aluminium coating is not
expected. The outward curvature of the Al3Ni towards the Al phase detected, furthermore, suggests diffusion
direction mainly from nickel or nickel rich phases towards aluminium [25]. The same phenomena can be
seen between the Al2Ni3 and the Al3Ni phases in Fig. 1. Because of a higher mobility of Al at 610°C,
14
opposite diffusion direction would be expected. In the work made by Jansssen and Rieck [24] only
aluminium was found to take part in the diffusion mechanism for Ni-Al couples at temperatures at about
600C. Wang et al. investigated Ni-Al couples prepared by rolling and annealing of Ni and Al sheets. After 5
and 15 minutes heat treatments at 650 C only Al3Ni and Al3Ni2 intermetallic phases were found. Both Ni
and Al were determined to be active diffusants at the annealing temperature.
Grain boundaries are known to provide high diffusivity paths in metals and diffusion along grain boundaries
is in an order of magnitude faster than bulk diffusion [33]. Identifying diffusion only or mainly in the grain
boundaries after short time of heat treatment, before the volume diffusion and leakage of diffusant through
the walls of the boundary into the adjoining crystals takes place, is therefore not surprising. The fact that it is
nickel atoms that diffuse into the grain boundaries of aluminium, instead of the opposite, is however
unexpected.
Harrison´s classification of diffusion kinetics [34] is the first and still the most common method applied to
explain the possible diffusion behaviour along grain boundaries. Harrison divided the diffusion mechanism
into three regimes called A, B and C kinetics, see Fig 12.

Fig. 12: Schematic illustration of Harrison´s three regime classification of diffusion kinetics. Where Dv is the volume diffusion
coefficient, Db is the grain boundary coefficient, t is the diffusion time, d is the spacing between the grain boundaries and δ is the
grain boundary width.
Regime A takes place under the conditions of high temperature and/or very long heat treatment and/or small
grain sizes. Under these conditions the volume diffusion length (Dvt)1/2 , where Dv is the volume diffusion
coefficient and t is the diffusion time, is much larger than the spacing between the grain boundaries (d).
Under these conditions leakage fields from each grain boundary overlap each other. Hence, the system
appears to obey Fick´s law of diffusion where the whole system has the same diffusion coefficient Deff. At
lower temperatures, and/or shorter diffusion time, and/or for polycrystals with larger grain size the diffusion
can be characterised under the B type regime. Still the grain boundary diffusion takes place simultaneously
with the volume diffusion. The difference is, that here the overlapping between leakage fields of each grain
is not actual. For regime C, diffusion may be considered to take place only within the grain boundaries. Here
the conditions are a lower temperature and/or shorter diffusion time and or when the volume diffusion
coefficient is much smaller than the grain boundary diffusion coefficient.
The diffusion kinetics during the first minutes of heat treatment for the PVD Al-Ni diffusion couples can be
described by Type C regime in Harrison´s diffusion system. After formation of the relatively small
crystalline Al3Ni phase, Type B and A become more and more dominant as recognized by the 10 and 20 min.
heat treatments. For longer heat treatments the diffusion kinetics obey primarily the Type A regime. During
the heat treatment a competition between the formation of Al3Ni and Al3Ni2 take place. From the
experiments of Castleman and Seigle [28] the diffusion coefficient of Al3Ni and Al3Ni2 are 1.8 x 10-11 and
9.1 x 10-10 cm2/sec, respectively. Hence, the Al3Ni2 phase grows faster than the Al3Ni phase and therefore,
for structures with limited Al sources, when heat treated for long enough time the Al-Ni diffusion film will
consist solely of the Al3Ni2 phase as evident from Fig. 1.
When the entire Al layer is consumed, somewhere in-between 30 min. and 2, hours AlNi and AlNi3 phases
start to form in between the Al3Ni2 phase and the Ni substrate. Further heat treatment results in slow growth
of these two phases and crystal growth of the Al3Ni2 phase. Simultaneously, a porous -Al2O3 phase is
formed on the top surface of the Al3Ni2. Local TEM analyses on a relatively large cross section area of 2
hours heat treated sample (Fig. 9) indicate no traces of AlNi3 in the Al3Ni2 intermetallic phase. From the
EBSD IQ map in Fig. 8, elongated grains of the Al3Ni2 structure are detected closest to the -Al2O3 phase.
The reason for the elongated grains can be slow diffusion kinetics due to a shortage of Ni reactants.
Correlating the present findings to the literature, the following can be found: Janssen´s and Rieck´s [24]
annealing experiments on Ni-Al couples for up to 66 hours resulted in formation of no other phases except
Al3Ni and Al3Ni2. Castleman and Seigle [28], however, showed that after all heat treatment at 600C for 340
hours AlNi3 and AlNi formed in between the Al3Ni2 and Ni phase. Tarento and Blaise [35] studied
interdiffusion between single crystal Ni substrate and evaporated 200 nm thick Al heat treated at 220 C for
up to 11 hours. They observed the AlNi3 phase to be the first to be formed after only a few minutes of heat
treatment. They explain the formation to be due to a lowering of stoichiometry of 2% of aluminium
concentration resulting in easy nucleation. In this study it is propose that the initial diffusion mechanism is
through movement of Al into Ni which results in formation of AlNi3. The overall diffusion mechanism was,
however, explained by manly being in the form of grain boundary diffusion. According to a review of thin
film aluminide formation by Colcan (1990) [36], predicting what phase will form initially during thin film
reaction is not yet possible. In another publication [37] the same author together with Mayer suggests that it
is the most aluminium rich phase in the phase diagram that forms first and that aluminium is the dominant
diffusion species during the initial phase formation.
Venezia et al. [38] reported formation of only -Al2O3 on the top of AlNi3 alloy exposed to 2.5e-7 oxygen
pressure and 700C temperature. It has also been shown that -Al2O3 can be formed by the annealing of
boehmite and pseudoboehmite at temperatures from 500-700C [39,40].
16
5 Conclusions
In the present study it has been shown that diffusion coupling of columnar structured PVD Al and large
crystalline Ni plate at 610C, the dominant diffusion mechanism is grain boundary diffusion of high Ni
containing phases into Al and Al-rich phases. The initial phase formed during the diffusion process is
proposed to be the AlNi3 phase. In addition to the formation of Al-Ni intermetallic alloys, -Al2O3 is detected
on the top surface of the samples after 2 hours of heat treatment. In Fig. 13 the composition of the PVD Al-
Ni diffusion couples as deposited, heat treated for 10, and heat treated for 120 minutes are schematically
illustrated.
Fig: 13: Schematic diagrams showing the metallic and intermetallic phases found in the as deposited, 10 min. and 120 min. heat
treated samples.
The main findings in this study are as follows:
As deposited stage
 The as deposited PVD Al layer has a columnar structure
 No epitaxial relation between Ni substrate and Al deposit is detected
10 min. of heat treatment at 610 C

 Very fine crystallites with unknown chemical composition are present in the Al3Ni2 layer
 The Al3Ni and Al3Ni2 diffusion layers have outward curvature towards the Al has inward curvature
strongly suggesting that the main diffusion is from higher Ni containing species to the lower Ni
containing species
 AlNi3 phase is formed on the grain boundaries and triple lines of Al columnar grains
 The grain size of the interdiffused layers, Al3Ni and Al3Ni2, are too small for EBSD analyses to be
prepared
120 min. heat treatment
 The average grain size of the AlNi3 layer closest to the Ni substrate is about 300 nm.
 The average grain size of the AlNi and Al3Ni2 is 812 nm.
 The Al3Ni2 microstructure is characterised with equi-axed grains and slightly elongated grains
closest to the top
 Relatively large area of the Al3Ni2 layer is investigated with high resolution TEM and only Al3Ni2 is
found
 Highly porous -Al2O3 is formed on the top surface of the diffusion layer.
Acknowledgement
The authors would like to thank Lars Pleth Nielsen and Kristian Rechendorff at The Danish Technological
Institute for assisting with the PVD coatings. Ewa Adamsen, Lars Pedersen, John C. Troelsen and Steffen S.
Munch at DTU are all acknowledged for their invaluable help. Financial support from The Energy
Technology Development and Demonstration Program in Denmark (EUDP) (project number: 63011-0200) is
gratefully acknowledged. The A.P. Møller and Chastine Mc-Kinney Møller Foundation is gratefully
acknowledged for their contribution toward the establishment of the Center for Electron Nanoscopy in the
Technical University of Denmark.
References
[1] Kjartansdóttir CK, Nielsen LP, Møller P. Development of durable and efficient electrodes for large-
scale alkaline water electrolysis. Int J Hydrog Energy 2013;38:8221–31.
[2] Kellenberger A, Vaszilcsin N, Brandl W, Duteanu N. Kinetics of hydrogen evolution reaction on
skeleton nickel and nickel-titanium electrodes obtained by thermal arc spraying technique. Int J
Hydrog Energy 2007;32:3258–65.
[3] Birry L, Lasia A. Studies of the hydrogen evolution reaction on Raney nickel-molybdenum
electrodes. J Appl Electrochem 2004;34:735–49.
[4] Kellenberger A, Vaszilcsin N. The determination of the roughness factor of skeleton nickel electrodes
by cyclic voltammetry. Rev Chim 2005;56:712–5.
[5] Boruciński T, Rausch S, Wendt H. Raney nickel activated H2-cathodes Part II: Correlation of
morphology and effective catalytic activity of Raney-nickel coated cathodes. J Appl Electrochem
1992;22:1031–8.
[6] Crnkovic F., Machado SA., Avaca L. Electrochemical and morphological studies of electrodeposited
Ni–Fe–Mo–Zn alloys tailored for water electrolysis. Int J Hydrog Energy 2004;29:249–54.
[7] Sheela G. Zinc-nickel alloy electrodeposits for water electrolysis. Int J Hydrog Energy 2002;27:627–
33.
[8] Hu WK. Electrocatalytic properties of new electrocatalysts for hydrogen evolution in alkaline water
electrolysis. Int J Hydrog Energy 2000;25:111–8.
[9] Los P. Hydrogen evolution reaction on Ni-Al electrodes. J Appl Electrochem 1993;23:135–40.
[10] Raj IA. Nickel-based, binary-composite electrocatalysts for the cathodes in the energy-efficient
industrial production of hydrogen from alkaline-water electrolytic cells. J Mater Sci 1993;28:4375–82.
[11] Wendt H, Imarisio G. Nine years of research and development on advanced water electrolysis. A
review of the research programme of the Commission of the European Communities. J Appl
Electrochem 1988;18:1–14.
[12] Miao HJ, Piron DL. Composite-coating electrodes for hydrogen evolution reaction. Electrochimica
Acta 1993;38:1079–85.
18
[13] Endoh E, Otouma H, Morimoto T, Oda Y. New Raney nickel composite-coated electrode for hydrogen
evolution. Int J Hydrog Energy 1987;12:473–9.
[14] Endoh E, Otouma H, Morimoto T. Advanced low hydrogen overvoltage cathode for chlor-alkali
electrolysis cells. Int J Hydrog Energy 1988;13:207–13.
[15] Choquette Y, Ménard H, Brossard L. Hydrogen discharge on a Raney nickel composite-coated
electrode. Int J Hydrog Energy 1989;14:637–42.
[16] Choquette Y, Brossard L, Lasia A, Menard H. Study of the Kinetics of Hydrogen Evolution Reaction
on Raney Nickel Composite‐Coated Electrode by AC Impedance Technique. J Electrochem Soc
1990;137:1723–30.
[17] Hitz C, Lasia A. Experimental study and modeling of impedance of the her on porous Ni electrodes. J
Electroanal Chem 2001;500:213–22.
[18] Chen L, Lasia A. Study of the Kinetics of Hydrogen Evolution Reaction on Nickel‐Zinc Alloy
Electrodes. J Electrochem Soc 1991;138:3321–8.
[19] Balej J, Divisek J, Schmitz H, Mergel J. Preparation and properties of raney-nickel electrodes on Ni-
Zn base for H2 and O2 evolution from alkaline-solutions. Part I: electrodeposition of Ni-Zn alloys from
chloride solutions. J Appl Electrochem 1992;22:705–10.
[20] Schiller G, Henne R, Borck V. Vacuum plasma spraying of high-performance electrodes for alkaline
water electrolysis. J Therm Spray Technol 1995;4:185–94.
[21] Miousse D, Lasia A, Borck V. Hydrogen evolution reaction on Ni-Al-Mo and Ni-Al electrodes
prepared by low pressure plasma spraying. J Appl Electrochem 1995;25:592–602.
[22] Fournier J, Miousse D, Legoux JG. Wire-arc sprayed nickel based coating for hydrogen evolution
reaction in alkaline solutions. Int J Hydrog Energy 1999;24:519–28.
[23] Singleton MF, Murray JL, Nash P. Al-Ni (Aluminium-Nickel). In: Massalski TB, Okamoto H,
Subramanian PR, Kacprzak L, editors. Bin. Alloy Phase Diagr., vol. 1, American Society for Metals;
1986, p. 142.
[24] Janssen M, Rieck G. Reaction diffusion and kirkendall-effect in nickel aluminium system. Trans
Metall Soc AIME 1967;239:1372–85.
[25] Konieczny M, Mola R, Thomas P, Kopciał M. Processing, Microstructure and Properties of Laminated
Ni-Intermetallic Composites Synthesised Using Ni Sheets and Al Foils. Arch Metall Mater
2011;56:693–702.
[26] Wang QW, Fan GH, Geng L, Zhang J, Zhang YZ, Cui XP. Formation of intermetallic compound layer
in multi-laminated Ni–(TiB2/Al) composite sheets during annealing treatment. Micron 2013;45:150–4.
[27] Tsao C-L, Chen S-W. Interfacial reactions in the liquid diffusion couples of Mg/Ni, Al/Ni and Al/(Ni)-
Al2O3 systems. J Mater Sci 1995;30:5215–22.
[28] Castleman LS, Seigle LL. Layer growth during interdiffusion in aluminum-nickel alloy system. Metall
Soc Am Inst Min Metall Pet Eng -- Trans 1958;212:589–96.
[29] Mackenzie JK. Second Paper on Statistics Associated with the Random Disorientation of Cubes.
Biometrika 1958;45:229–40.
[30] Thornton JA. The microstructure of sputter‐deposited coatings. J Vac Sci Technol A 1986;4:3059–65.
[31] Kelly P., Arnell R. Magnetron sputtering: a review of recent developments and applications. Vacuum
2000;56:159–72.
[32] Pauleau Y. Generation and evolution of residual stresses in physical vapour-deposited thin films.
Vacuum 2001;61:175–81.
[33] Inderjeet Kaur, Mishin Y, Gust W. Fundamentals of grain and interphase boundary diffusion.
Chichester; New York: John Wiley; 1995.
[34] Harrison LG. Influence of dislocations on diffusion kinetics in solids with particular reference to the
alkali halides. Trans Faraday Soc 1961;57:1191–9.
[35] Tarento RJ, Blaise G. Studies of the first steps of thin film interdiffusion in the Al-Ni system. Acta
Metall 1989;37:2305–12.
[36] Colgan EG. A review of thin-film aluminide formation. Mater Sci Rep 1990;5:1–44.
[37] Colgan EG, Mayer JW. Aluminium-Transition Metal Thin-Film Reactions. MRS Online Proc Libr
1988;119:null–null.
[38] Venezia AM, Loxton CM. Low pressure oxidation of Ni3Al alloys at elevated temperatures as studied
by x-ray photoelectron spectroscopy and Auger spectroscopy. Surf Sci 1988;194:136–48.
[39] Han L, Jun-Qiao W, Ji-Zhou C. Structure imperfection of γ-Al2O3. Polyhedron 1995;14:445–9.
[40] Lee M-H, Cheng C-F, Heine V, Klinowski J. Distribution of tetrahedral and octahedral A1 sites in
gamma alumina. Chem Phys Lett 1997;265:673–6.
Appended Paper IV 1

Unveiling the secrets of the Standard Hydrogen Electrode

- An inspiration for the on-going development
of hydrogen electrocatalysts
Martin Flyvbjerg(a), Cecilía K. Kjartansdóttir(a), Per Møller(a) & Michael Caspersen(a,b)
a) Department of Mechanical Engineering, Technical University of Denmark (DTU), Materials and Surface Engineering Sec‐
tion (MTU), Produktionstorvet Building 425, 2800 Kgs. Lyngby Denmark.
b) Siemens A/S, Borupvang 9, 2750 Ballerup
KEYWORDS Platinum, hydrogen evolution reaction, electrochemical potential, catalysis, surface area

ABSTRACT: New aspects in the on‐going discussion of what contributes to the electrode potential during hydrogen
evolution reaction (HER) are put forward. The focus is on available number of active sites vs. intrinsic material properties,
unveiling for the first time the nano‐scale topography of the electrode. The Platinum Black electrode has always be known
to have a very well defined potential used as reference for the electrochemical scale, and the particular surface topography
is basis for inspiration for electrochemical catalysis and other processes like hydrogenation where other catalysts are
used, such as Raney nickel. A new perspective on electrode design and electrodeposition of the platinum black electrode
are presented.

bridge; radiation hitting the strip will cause a rise in tem‐
Background perature and thereby electrical resistance. Prior to the
The platinum black electrode is accepted worldwide as platinum black experiments, a metallic strip blackened by
the Standard Hydrogen Electrode (SHE) and in electro‐ soot from a burning petroleum flame was used, however
chemistry it is defined the reference electrode for the the exact layer of soot was not easy to reproduce.
absolute zero potential. Platinum is known for good cor‐ It was acknowledged that a layer of completely black
rosive resistance, extremely good catalytic properties, platinum could be electrodeposited from a (platinum
preferred electrocatalysis in fuel cell and environmental chlorid) “platinchlorid” bath by adding copper or lead to
industry1–3, and may serve as solution as energy storage the bath. The actual composition based on hexachloro‐
for sustainable energy sources, known as the “hydrogen platinic acid or platinum(IV) chloride. Kohlrausch point‐
economy”4,5. ed out that this term is commonly applied to hexachloro‐
Smooth platinum comprises greater hydrogen overvolt‐ platinic acid. Adding copper sulphate to the plating bath
age, than platinized platinum (platinum black) 150mV vs. was found to give larger, smoother and less brittle plati‐
0 at 1 mA/cm2 6. Therefore one would assume that plati‐ num black coating. However, adding a very small amount
num black electrode is the ideal electrode for water elec‐ of lead acetate instead yielded even better results. Later
trolysis as well as other reactions that depend upon low on other additives have been added to improve the adhe‐
hydrogen overpotential. A very important question arises sion between the platinum layer and the substrate. In‐
what actually contributes to the very low hydrogen over‐ spired by Arrhenius' dissociation theory, Nobel laureate
voltage? Is overvoltage a material constant or does the Walter Nernst89 developed his theories of thermodynamic
microstructure influence the catalytic behaviour? aspects of an electrochemical cell in 1889, presently
known as the Nernst equation. The theory provides a
The historical origins relation between the hydrogen and hydroxyl ion concen‐
The history of platinum black goes as far back as 1890 tration.
where O.R. Lummer and F. Kurlbaum 7 started searching In 1893 Max Le Blanc 10 made a practical discovery in an
for a black substrate to replace the current method for attempt to measure the hydrogen ion concentration in a
measuring radiation, by creating a reproducible bolome‐ given solution, today described as pH, by letting a stream
ter for measuring the black body radiation. of hydrogen gas flow around a platinum electrode, elec‐
Blackbody radiation was at that time measuring the re‐ troplated with platinum (platinized platinum/platinum
sistance though metal strip connected to a Wheatstone
2
black) the eleectrode will beehave like a hyydrogen electtrode reaaction is solelyy a chemical r
reaction. The table below
i.e. absolute p
potential. shoows the Volm mer, Heyrovskyy and Tafel reactions as theey
Thereby hee established a
a basis for the electromotive tak
ke place in aciidic and alkaliine solutions.
force measureement in a celll, by the know
wn hydrogen half‐
cell redox reaaction:
2 2 →
According tto Trasatti 11 tthe SHE is a sccale for obtainning
the absolute e electrode poteential, definedd as the differeence V
Volmer, Heyrovsky reacttions:
in electronic e
energy betweeen an electrollyte and the Fermi Acidic Alk
kaline
energy of a m
metal electrodee. Fermi level is defined as t the
V
Volmer reac‐ →
energy level b
between a poin nt in a solid c
crystalline and
d an tio
on →
electron at rest in vacuum..
H
Heyrovsky H O
In order to match the ph hysical scale to
o the electroch
hem‐ reaaction → →
ical scale the SHE is defineed by IUPAC11 via following
relation at 298.15 K: T
Tafel → →

. D
During hydro ogen evolution n the two discharging steps
occcur simultaneeously, the sloower step deteermines the
Where EM(aabs) is the abssolute potentiial for the mettal M HE ER rate. From the previous it is clear the activation
and SHE the h
hydrogen elecctrode. The orrigin of the eleec‐ en
nergy decreasees with increassed adsorption n energy (M–H
trochemical s
scale predates the physical s scale, and is d
de‐ boond strength), while increassed adsorption n energy mean ns
termined by a
a redox half‐ceell reaction. inccrease in term
ms of H coveerage on the e electrode sur‐
2 2 → facce. Therefore i
if they M—H bond energy is too strong
This reactioon occurs at aa platinized platinum electrrode, thee H will occcupy the availlable surface s sites and inhib
bit
submerged in n an acidic sollution with hyydrogen bubblling thee second step of the total reeaction. Thus the best hy‐
thought it. Th he relation and theory betw ween electrodee and droogen electrodde should be th he one having g intermediatee
electrochemiccal potential i bed in 12,13. Thee
is fully describ M––H bond enerrgy (or free en nergy of hydro ogen adsorptio on
SHE is occasionally confused with the N Normal Hydrog gen (ΔGG )), as statted in the Sabbatier principlee15. When plott‐
Electrode (NH HE). The NHE E is defined ass the potentiall of tin
ng the electroccatalytic activvity (exchangee current denssi‐
platinum in a a 1N acid soluttion 13 and wass used as referrence ty for HER) vs. t
the M–H bond d strength forr different mett‐
electrode in t he early days of electrochem mistry. The SH HE alss a so called vo
olcano plot is formed. The v volcano plot i
in
is on the otheer hand defineed by the conccentration of H H+ Figgure 1 supportts Sabatier’s thheory and shoows clearly thaat
in the electrode/solution in the SHE the ac‐
nterface. For t plaatinum should d be the most active metal for hydrogen
tivity of the h
hydrogen half‐‐cell reaction m must be of aH H+ = evo olution.
1, as derived f
from Nernst equation, undeer standard co ondi‐
tions.

The electro
ocatalytic effe
ect of a hydro
ogen electrod
de
The ability to adsorb hyd drogen atoms plays a key ro ole in
the mechanissm and kineticcs of hydrogen n electrodes. S
Sev‐
eral transitionn metals have such a strong g M – H bond that
they are able to dissociate h hydrogen molecules in a so olu‐
tion, this is paarticularly thee case for Pt. H
Hence the Pt –– H
bond is strong ger than the H H – H bond 14..
The hydrog gen adsorption
n mechanism on a metal (M
M)
surface is com
mmonly writteen as
→
Most theories state that the adsorbed hydrogen ato oms
combine into o hydrogen mo olecules eitherr by reacting w
with
further dischaarging H+ or b
by recombinin ng with another
adsorbed hyd drogen atom. The first disch
T harge step wh here
hydrogen is a
adsorbed at th
he electrode su urface is knowwn as F
Figure 1: The d
dependence of
f the electrocattalytic activity for
14
4
the Volmer reeaction, and th he second steep where hydrrogen HE
ER on the metaal – hydrogen b
bond formed .
molecules aree formed is knnown as the H Heyrovsky reacc‐ A more recennt way to defin ne the volcano o curves princci‐
tion. The recoombination off two adsorbed hydrogen attoms plee is based on t
the electronicc configuration n of the atomss
is known as TTafel reactionss. Both Volmeer and Heyrovvsky witthin the latticce of the catalyyst material16––19 or the so
reactions are electrochemical reactions w whereas the T
Tafel callled hypo‐ hyp per‐d theory. Here metals o on the left side

Appended Paper IV 3

of the volcano plot are called hypo‐d‐electronic metals In the latter case the conclusion has been made that
because they have empty or half‐filled vacant d‐orbitals higher activity stems from creation of new lattice vacan‐
and the metals on the right side of the volcano plot is cies along with an increase in surface area is obtained
called hyper‐d‐ electronic elements because they have during leaching 27. In fact, several authors have ascribed
internally pared d‐electrons which are not available for increased efficiencies of alloys compared to plain sub‐
bonding in pure metals. The optimum (best catalyst, the stances, often nickel, to gain in surface area rather than
catalyst at the top of the volcano plot) is either defined20 intrinsic factors28,29. A larger surface area per unit mass
to be at d8 or d5. will all things being equal increase the amount of active
From the volcano plot one can imagine that the elec‐ surface sites thus lowering the local current density and
trocatalytic activity towards hydrogen evolution could be reduce the required potential for the reaction to proceed,
tailored by combining metals from the left‐hand side of but this does not directly prove or disprove the d‐electron
the volcano plot (hypo‐d‐electron metals) with metals theory. Indeed, it has been shown29–31 how surface area
from the right‐hand side of the volcano plot (hyper‐d‐ effects can be combined with electronic effects from vari‐
electron metals). This seems to be the case and it has ous noble metals to reach higher catalytic efficiency for
been shown that the activity of intermetallic phases and the HER.
alloys of transition metals towards the HER obey the This can be supported by work done on particular fa‐
same sort of volcano plots as pure metals21. vourable crystallographic orientations of platinum in
Recently, J. Greeley et al.22 used the density function neighbouring studies32, implying that higher order index
theory (DFT) to calculate the free energy of hydrogen facets are more active in nature due to higher number of
adsorption for over 700 binary transition‐metal surface steps and missing atomic bonds (high surface energy).
alloys. Their calculations showed that the electrocatalytic Equivalence back to the high activity of leached alloys
activity of BiPt towards HER is comparable, or better, seems reasonable. Here we argue that a complex correla‐
than for pure Pt. Electrochemical testing on a synthesized tion must exist between the various contributions and
BiPt alloy was showed to support the theoretical evalua‐ that all contributions should be tailored in order to max‐
tion. imize electrocatalytic activity.
Although illuminating some interesting trends there
are some uncertainties to whether DFT calculations by
their own are viable for actual electrocatalytic surfaces. In Experimental procedure
its nature DFT is considering only the ground state of a Hexachloroplatinic acid H2PtCl6 for the plating bath
material in order to evaluate its pure intrinsic properties. was prepared by dissolving a wire of pure platinum in
This approach will though, in spite of delivering interest‐ aqua regia, (1 part of concentrated nitric acid, and 3 parts
ing theoretical predictions, not necessarily give the right of concentrated hydrochloric acid). In order to denox the
indication for actual electrocatalytic surfaces where crys‐ solution, additional HCl 20 mL was added three times
tallographic defects and lattice distortion will alter the and boiled off, to ensure the solution was free of any addi‐
system away from its ground state. Furthermore, often tional HNO3 and NO2. The solution was boiled down and
limited number of atoms constitutes the idealized single DI‐water was added to reach the desired concentration of
crystal considered in the calculations due to the available 0.072M (3.5 %) H2PtCl6 with pH of 0.8.
computer power. Hence, other bulk phenomena occur‐ According to the original recipe for platinum black a
ring in a solid polycrystalline specimen such as interac‐ piece of platinum foil is desired for the platinization pro‐
tions from grain boundaries, segregation and atomic cess. However due to the current price of Pt a gold plated
impurities are neither taken into consideration. substrate is used as replacement. The Au is selected to
As a consequence, consistency with actual observations is ensure good adhesion. Several authors have reported
not always obtainable, and indeed the practical usage of adhesion problems in the production of platinized plati‐
DFT has yet to be further evaluated and documented, num when using additive free plating solutions 33–36. The
until it can potentially be used as an impactful argument. lattice constant of Au is relatively close to the lattice con‐
For genuine optimization of electrocatalytic surfaces the stant of Pt, as the original recipe of platinized platinum
above mentioned phenomena has to be emphasized as requires. For the electrode substrate, a piece of stainless
well as how they will affect the final electrocatalyst. steel AISI 304 10 x 10 x 1 mm was selected, initially plated
The actual electrocatalytic effect does not only depend with strike nickel from Wood’s electrolyte followed by a
on the electron configuration. The structure and topogra‐ layer of sulphamate Ni (~15µm). The substrate was there‐
phy of the catalyst also has a great influence of the appar‐ after plated with 1.5 µm gold layer.
ent electrocatalytic efficiency. The platinum was deposited, with current density of 3
Previous studies have shown that improved efficiency, A/dm2, from the hexachloroplatinic acid with 1.3 * 10‐4 M
lower cell potential, in e.g. PEM fuel cells or electrolysis (0.005 %) Pb‐acetate trihydrate added to the electrolyte.
cells can be obtained by modifying the electrode mor‐ The Pb is addet in order to achieve the deep black surface
phology with various techniques. Depositing active layers as described in literature 37. The developed electrode was
on highly porous carbon supports23,24 or creating highly cleaned by gentle dipping in DI‐water and dried prior to
porous electrodes prepared by selectively leaching of one any investigations. However, it should be mentioned that
or more elements from metal alloys are good examples of the original recipe states the electrode must be kept wet
this25,26. and stored in DI‐water for maximum catalytic properties.

4
In order to estimate the surface area of the Pb‐black
electrode Brunauer‐Emmett‐Teller (BET) method was
employed. The Pt‐black electrode for the BET analyses
was prepared on a copper wire followed by a nickel and
gold layer as described earlier. The analysis was carried
out on a Micromeritics ASAP 2020, using N2 at liquid
nitrogen temperature. Prior to the measurements the
sample was degassed at 200°C in vacuum for 6 hours. A
commercial nickel foam sample was also investigated
using same procedure for comparison.
The structure and morphology of the Pt electrode was
investigated using a FEI Helios Nanolab 600i, a Field
Emission Gun‐Scanning Electron Microscope (FEG‐SEM)
at DTU‐Cen (Center for Electron Nanoscopy). The maxi‐
mum resolution of the SEM is listed to be better than 1.5
nm at 1kV.

Appended Paaper IV 5

Figure 2 FEG
G‐SEM microg graph of Pt‐blaack electrode a
as de‐ F
Figure 4 FEG‐SEM micrograaph of Pt Black
k electrode, w
with
posited, show
wing a flowerr‐like morpho
ology (full win
ndow inccreased magniffication (x 350
0.000).
magnification x15.000).

F
Figure 5 FEG‐S SEM micrograaph of Pt Black k electrode, wwith
thee highest reached magnificaation. Small w
white dots appeear
Figure 3 FE
EG‐SEM microograph of Pt‐BBlack electrodee, possibly being neew nucleationss. (x 500.000).

w
with increased n (x 50.000).
d magnification

6
Tailoring the optimal surface mo

orphology nd Pt(II)Cl42‐ ) is reduced to pure Pt. How
an wever this theoo‐
ry does not expllain the role o
of lead or the c
chemistry in‐
For any elecctro depositedd materials th
he structure off the
volved.
layer dependss upon two crucial factors ‐‐ the current d den‐
sity and the in
nhibition of d
deposition. Thhere are in gen neral O
One of the claaimed reaction
n couples in t
the electrodep
po‐
two theories w we can apply to the electrodepostion of P Pt sition of platinu
um from chlorroplatinic acid
d is:
black, either P
Pb can act as an acceleratorr for the platinng Pt IV Cl 2e ⇄ PtCl 2
2Cl
process or it c
can inhibit thee formation o
of hydrogen att the T
Thereafter plaatinum is deposited according to:
cathode thereeby allowing P Pt to be depossited at greateer
rate.
PtCl 2e → Pt 4Cl
René Winand created a d diagram, Figu ure 7 showing the
8
86.35 kcal
ayers made wiith electro dep
structure of la position as a t
tool
for classifyingg deposited layyers. By using
g the Winand and
the Pourbaix diagrams for platinum and d lead as tools it T
The Gibbs free energy (ΔG)) for depositio
on is negative
might be posssible to work out the theoriies of platinum m ind
dicating that t
the reaction is
s thermodynaamically favou
ur‐
black plating.. able.
One theory 400 states possible adsorption of lead wheree,
O
thee adsorbed Pbb ions act as siites for electro
on transfer
between the surrface and the P PtCl62‐ ions. TThese Pb ions
couuld be considered “pseudo‐‐defects” actin ng as nucleation

Figure 7 A s simplified version of the diag gram of R. Win nand,
showing differrent types of polycrystalline e electrodepositts as a
+ 338
function of J //CMez (or J / Jd) and inhibiition intensity . FI:
F
Figure 6 Pourb
baix diagram ffor Pt – Cl an
nd Pb with re‐‐
Field oriented d isolated crysttals, BR: Basis oriented repro oduc‐
bath concentraation.
speective plating b
tion, Field oriented texture type, UD: Uno oriented dispeersion
type, 2D: bidim mesional nuclleation, 3D: triidimesional nu uclea‐ cenntres for Pt isllands on the fl
flat areas. Oth
her experi‐
tion. meental results 411 indicate thatt “Pb‐acetate s
significantly
en
nhances the eleectrode reactiions in platinu um black coatt‐
By compariing the structu ure of the elecctrode with th he
ing
g by mainly lo owering the ennergy barrier f for the reduc‐‐
Winand diagrram the structture is likely t to be of BR or FI
tio
on of Pt (IV) to o Pt and by su
uppressing thee reduction off
type rather thhan UD or FT.. Both BR and d FI indicate lo ow
Pt (IV) to Pt (II))”.
current densiity and lack off inhibition. BBy adding Pb t to
the plating baath the surface structure wiill change dramat‐ H
However no s single elementt has the ability to suppresss
ically and theerefore the cheemistry of thee plating bath change in oxidaation states. Thhese theories simply canno
ot
would also beeen altered. Byy once again c consulting thee be supported byy any of the fo
ollowing chemmistry, or the
Winand diagrram the typicaal structure iss likely to be u unor‐ Po
ourbaix diagraams.
iented disperssive growth or the unnameed region just
above. In order to match th he diagram a p plausible theo ory is T
The Pourbaix diagram is a p
powerful tool l to show ther‐
that Pb slighttly inhibits thee deposition o
of Pt resulting g in mo odynamic stab ble complexess at various pH H and potentiial.
forced new nu ucleation, cauusing the dend drite‐like structure Ass shown in Figgure 6, Pt and the PtCl62‐ ion
n appears in t
the
to constantly spawn new b branches but o only allowing diaagram. More i interesting aree the diagramm for Pb Figuree
them to grow w to a specific size before a n
new branch iss 8, keeping in miind the pH of
f the plating ellectrolyte wass
spawned and the process is s repeated. aroound 0.8 two ions are noticceable PbCl‐ and PbCl4‐.
There have been multiplle theories forr how the plating
process takes place, the mo ost popular is likely describ
bed
by 39,40 where couples of plaatinum chloriide (Pt(IV)Cl622‐

Appended Paaper IV 7

Fig
gure 8 Pourbaixx diagram for Pb ‐Cl PbCl+ is especially notice‐
able s
species likely t
to influence thee plating of Pt‐‐Black.
One could imagine that Pb III Cl cou
uld be reducedd to
Pb II Cl , and
d thereby chan
nging the oxiddation state off Pb
and creates an
n initiation sttep for the plaatinum platingg.
During the prrocess Pb is reeduced from o oxidation statee +3,
to +2. As seen
n in following reaction the l lead ion can b
be
oxidized by th
he platinum a acid, leaving bbehind pure pllati‐
num. Calculatted at 20ºC.
4PbCl PtCl 6Cl → Pt 4PbCl
381.6 kcal
This makess the PbCl+ a r
reduction agen nt for the PtCll
ion, oxidizingg PbCl+ to 4PbbCl4‐. As menttioned earlier
according to 440 the platinumm plating relies only on red duc‐
tion of platinu
um chloride, w when calculatting the ΔG off
both reactionns the reactionn involving Pb bCl is much m more
plausible to h
happened rath her than simplle PtCl2 or PtC Cl4
complex redu uced to Pt. This could explaain the requiree‐
ment for Pb tto be present i in order to be able to electrro‐
deposit the pllatinum‐black since PtCl62‐ iss
k electrodes, s
instantaneoussly reduced an nd creates thee very delicatee
nanocrystals.
If any Pb wwas to be found d in the electrrodeposited laayer
of the electrode the overvo oltage of the electrode is no
ot
likely to be ass low, since , l
ead have an e exchange curreent6
of 2 ∙10‐13 A/cm
m2.
An alternattive plausible theory of a Pb
b free layer staates,
one could arggue lead couldd possibly as p
pure Pb, after‐‐
wards be ablee to reduce thee PtCl62‐ to Pt via the follow
wing
reactions.
4Pb
b 3PtCl → 3Pt 4PbCl 2Cl
or
2
2Pb PtCl → Pt PbCl 4Cl

8
Results and Discussion In spite of uncertainties connected to the use of this

method the magnitude of the found roughness factor
When looking at the deposited Pt‐black electrode with
clearly renders the uncertainties insignificant, and there
the bare eye the electrode appears completely black and
can be no doubt that the Pt black electrode possesses a
does not reflect any incident light, i.e. acts like a perfect
natural huge surface area.
black body, as original intended as described earlier.
The Pt‐black electrode can serve as inspiration for fab‐
FEG‐SEM micrographs of the Pt‐black electrode sur‐
ricating of HER catalysts. By mimicking the morphology
face, as deposited, are shown in Fig. 2 to 5. The micro‐
of the Pt‐black electrode with a cheaper material the
graphs reveal very distinct features that remarkable
electrocatalytic properties can be altered to more positive
resampling the flowerhead of a Chrysanthemum flower at
direction. Here the importance of large amount of active
a nano‐scale. With increased magnification it becomes
sites in the development of catalysts has been empha‐
clear that the pattern of growth repeats itself down to‐
sised. By combining the knowledge of how the electronic
wards nano‐scale level.
configuration of a catalyst influences catalytic properties
The surface of the electrode is very fragile and black (as explained with the Volcano plot) to the importance of
material flakes of if touched. This delicate surface topog‐ high active surface area less expensive and more active
raphy explains the pitch black appearance of the elec‐ HER catalysts can be produced.
trode. Hence all incident and diffuse light are trapped
Another well‐known catalyst metal with a similar struc‐
unable to escape the surface.
ture, Raney‐Nickel shares some of the properties with the
Various theories for plating platinum black have been Pt‐black electrode. Both materials have a large surface
discussed earlier in this article. Clearly lead plays an im‐ area with numerous intrinsic sites for reaction. It is no
portant role when depositing Pt‐black. In order to detect secret within the catalyst industry 30,31,43, that morpholo‐
if the supposed lead chloride complex reactions will take gies with high surface area increase the activity of a het‐
place an in‐situ ultraviolet analysis of the plating process erogeneous reaction.
is required. From a theoretical point of view this reaction
The volcano plot based upon the Sabatier principle is
could explain the requirement for lead to be present in
often calculated using the density functional theory
the solution in order to achieve the pitch black surface.
(DFT), a quantum mechanical mathematic model based
The pitch black platinum electrode proves to hide a upon energy levels. Then calculating these models often a
very unique crystal structure as revealed in the SEM mi‐ single crystal is considered or a system with very limited
crographs, a morphology with extremely large surface configuration limiting the use of the model. Since reac‐
area, which holds immense amount of intrinsic sites tions occur at the intrinsic sites along the edges of a crys‐
along the crystal edges where reactions such as the HER tal it is necessary to consider different spacious configura‐
is likely to occur. The combination of a large area with tions and consider all the intermetallic compounds be‐
lots of intrinsic sites is likely to result in good possibilities tween the elements. This does not consider where or not
for electrocatalytic reactions. Hence the morphology of the intermetallic is thermodynamically stable or possible
this ideal electrode evidently plays a crucial role for the to produce.
overvoltage.
This increases the difficulties related with create a
Deposition trials without the addition of Pb resulted in workable mathematical model for electrocatalytic design
flaking of the Au and Pt layer, a possible explanation for and one should question the usefulness of already exist‐
debonding is likely stress in the Pt layer peeling off both ing models.
layers. This was not investigated further the focus was
The plating process of Pt‐black electrodes occurs at
pointed towards electrolyte with lead.
such low potentials that hydrogen formation is impossi‐
From the BET analysis the surface area of the Pt‐black ble to avoid, despite the electrode is unable to function as
is estimated to have a “roughness factor” of approximately cathode for electrolysis of water. The electrode is purely
5900 times compared to a geometric smooth surface. The designed for measureing the reversible reaction of hydro‐
roughness factors of the Pt‐black electrode have been gen, measuring the potential where hydrogen gas is aer‐
reported in the range of 3900‐20000 42. However in this ated around the electrode in an acidic solution. If used as
case helium was used for the measurement instead of cathode for making hydrogen the fine structure and to‐
nitrogen as in the present study. Helium molecules are far pography of the electrode is likely to be destroyed due to
smaller than nitrogen and possibly able to enter smaller gas erosion.
cavities. By comparison to commercial available nickel
When designing electrocatalysts the mechanical and
foam sample with a roughness factor in the range of 300
thermodynamic stability must be taken into considera‐
times greater.
tion.
It must be kept in mind the BET analysis is sensitive

towards the weight of the sample and the thickness of
different layers applied during the plating process. The ACKNOWLEDGMENT
recipe for Pt‐black does not state how thick a layer will be Peter Jacob Schwencke Westermann is acknowledged for
applied or the efficiency of the plating bath, so the final Pt assisting in the plating process of the platinum
layer was estimated by subtracting the inner layers (de‐ Hossein Alimadadi is acknowledged for assisting in operat‐
termined by SEM), so that only the Pt layer density was ing the FEG‐SEM, sample preparation and great inspiration.
evaluated in the BET measurement

Appended Paper IV 9

REFERENCES materials for electrochemical hydrogen production.

INTERNATIONAL JOURNAL OF HYDROGEN ENERGY, 2011, 36,
(1) GRIGORIEV, S.; POREMBSKY, V.; FATEEV, V. 12079–12087.
Int. J. Hydrogen Energy 2006, 31, 171–175. (31) Dominguez‐Crespo, M. A.; Torres‐Huerta, A. M.;
(2) Granovskii, M.; Dincer, I.; Rosen, M. a. J. Power Sources Brachetti‐Sibaja, B.; Flores‐Vela, A. Electrochemical performance
2006, 157, 411–421. of Ni‐RE (RE = rare earth) as electrode material for hydrogen
(3) KREUTER, W. Int. J. Hydrogen Energy 1998, 23, 661– evolution reaction in alkaline medium. INTERNATIONAL
666. JOURNAL OF HYDROGEN ENERGY, 2011, 36, 135–151.
(4) Bockris, J. O. PROCEEDINGS‐ Electrochem. Soc. PV (32) Tian, N.; Zhou, Z.‐Y.; Sun, S.‐G.; Ding, Y.; Wang, Z. L.
2000, 2000‐20, 1 – 24. Sci. 2007, 316 , 732–735.
(5) Bockris, J. O.; Veziroglu, T. N. Int. J. Hydrogen Energy (33) Jones, G.; Bollinger, D. M. J. Am. Chem. Soc. 1935, 57,
2007, 32, 1605–1610. 280–284.
(6) Baboian, R. Corrosion tests and standards : application (34) Von Hevesy, G.; Somiya, T. Platinum black.
and interpretation; 2005. ZEITSCHRIFT FUR PHYSIKALISCHE CHEMIE‐ABTEILUNG A‐
(7) Lummer, O.; Kurlbaum, F. Ann. Phys. 1892, 282, 204– CHEMISCHE THERMODYNAMIK KINETIK ELEKTROCHEMIE
224. EIGENSCHAFTSLEHRE, 1934, 171, 41–48.
(8) Von Ettingshausen, A.; Nernst, W. Ann. Phys. 1888, (35) Popoff, S.; Kunz, A. H.; Snow, R. D. J. Phys. Chem. 1928,
269, 474–492. 32, 1056–1060.
(9) Nernst, W. Ann. Phys. 1887, 267, 760–789. (36) Joncich, M. J.; Hackerman, N. PREPARATION AND
(10) Le Blanc, M. Zeitschrift für Phys. Chemie 1893, 12, 333. SURFACE AREA MEASUREMENTS OF PLATINIZED‐
(11) Trasatti, S.; of Pure, I. U.; Division, A. C. P. C. The PLATINUM ELECTRODES. JOURNAL OF THE
Absolute Electrode Potential: An Explanatory Note ELECTROCHEMICAL SOCIETY, 1964, 111, 1286–1289.
(recommendations 1986); Blackwell Scientific, 1986. (37) Kohlrausch, F. Ann. der Phys. und Chemie 1897.
(12) Biegler, T.; Woods, R. J. Chem. Educ. 1973, 50, 604. (38) Winand, R. ELECTRODEPOSITION OF METALS AND
(13) Ramette, R. W. J. Chem. Educ. 1987, 64, 885. ALLOYS ‐ NEW RESULTS AND PERSPECTIVES.
(14) Trasatti, S. In Encyclopedia of Electrochemical Power ELECTROCHIMICA ACTA, 1994, 39, 1091–1105.
Sources; Jürgen Garche, Ed.; Amsterdam, 2009; pp. 41–18. (39) Feltham, A. M.; Spiro, M. PLATINIZED PLATINUM
(15) Sabatier, P. Berichte der Dtsch. Chem. Gesellschaft 1911, ELECTRODES. CHEMICAL REVIEWS, 1971, 71.
44, 1984–2001. (40) Layson, A. R.; Columbia, M. R. The morphology of
(16) Jaksic, J. M.; Krstajic, N. V.; Grgur, B. N.; Jaksic, M. M. platinum black electrodeposited on highly oriented pyrolytic
Int. J. Hydrogen Energy 23, 667–681. graphite studied with scanning electron microscopy and
(17) Jaksic, M. M. Int. J. Hydrogen Energy 2001, 26, 559–578. scanning tunneling microscopy. MICROCHEMICAL JOURNAL,
(18) Neophytides, S. G.; Zafeiratos, S.; Papakonstantinou, G. 1997, 56, 103–113.
D.; Jaksic, J. M.; Paloukis, F. E.; Jaksic, M. M. Extended Brewer (41) Saitou, M. Electrochemical characterization of
hypo‐hyper‐d‐interionic bonding theory ‐ I. Theoretical platinum black electrodeposited from electrolyte including lead
considerations and examples for its experimental confirmation. acetate trihydrate. Surface and Coatings Technology, 2007, 201.
INTERNATIONAL JOURNAL OF HYDROGEN ENERGY, 2005, 30, (42) Brodd, R. J.; Hackerman, N. POLARIZATION
131–147. CAPACITY AT SOLID ELECTRODES AND TRUE SURFACE
(19) Jaksic, M. M. ELECTROCATALYSIS OF HYDROGEN AREA VALUES. JOURNAL OF THE ELECTROCHEMICAL
EVOLUTION IN THE LIGHT OF THE BREWER‐ENGEL SOCIETY, 1957, 104, 704–709.
THEORY FOR BONDING IN METALS AND INTERMETALLIC (43) Rosalbino, F.; Maccio, D.; Angelini, E.; Saccone, A.;
PHASES. ELECTROCHIMICA ACTA, 1984, 29, 1539–1550. Delfino, S. Characterization of Fe‐Zn‐R (R = rare earth metal)
(20) Jaksic, M. M.; Lacnjevac, C. M.; Grgur, B. N.; Krstajic, crystalline alloys as electrocatalysts for hydrogen evolution.
N. V. Volcano plots along intermetallic hypo‐hyper‐d‐electronic INTERNATIONAL JOURNAL OF HYDROGEN ENERGY, 2008, 33,
phase diagrams and electrocatalysis for hydrogen electrode 2660–2667.
reactions. JOURNAL OF NEW MATERIALS FOR
ELECTROCHEMICAL SYSTEMS, 2000, 3, 169–182.
(21) JAKI, J. Int. J. Hydrogen Energy 1998, 23, 667–681.
(22) Greeley, J.; Jaramillo, T. F.; Bonde, J.; Chorkendorff, I.
B.; Nørskov, J. K. Nat. Mater. 2006, 5, 909–13.
(23) Popovski, O. NEW TRENDS Res. Energ. Mater. 2006,
416 – 422.
(24) Paunovic, P. Maced. J. Chem. Chem. Eng. 2007, 26, 87 –
93.
(25) Brennecke, P. W.; Ewe, H. H. Energy Convers. Manag.
1991, 31, 585–594.
(26) Choudhary, V. R. J. Chem. Technol. Biotechnol. Chem.
Technol. 1983, 33A, 339 – 349.
(27) Bagotsky, V. S. Fuel cells : problems and solutions;
Wiley, 2012; p. 1 vol. (XVIII–385 p.).
(28) Birry, L.; Lasia, A. NEW Mater. Electrochem. Syst. 2001,
69–72.
(29) Herraiz‐Cardona, I.; Ortega, E.; Vazquez‐Gomez, L.;
Perez‐Herranz, V. Electrochemical characterization of a
NiCo/Zn cathode for hydrogen generation. INTERNATIONAL
JOURNAL OF HYDROGEN ENERGY, 2011, 36, 11578–11587.
(30) Solmaz, R.; Kardas, G. Fabrication and characterization
of NiCoZn‐M (M: Ag, Pd and Pt) electrocatalysts as cathode

Appendix
Appendix
GI-X-ray diffractograms for the PVD Al-Ni couples after 24 hours of heat
treatments, before and after leaching.() Al, () Al3Ni2. The incident angel is 6
degrees for both specimens.

Development of Hydrogen Electrodes For Alkaline Water Electrolysis

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Development of Hydrogen Electrodes For Alkaline Water Electrolysis

Diunggah oleh

Hak Cipta:

Format Tersedia

Downloaded from orbit.dtu.

dk on: Jan 30, 2018

Development of Hydrogen Electrodes for Alkaline Water Electrolysis

Kjartansdóttir, Cecilía Kristín; Møller, Per

Link back to DTU Orbit

Kgs. Lyngby, 3 December 2013

aluminising of nickel followed by thermo-chemical diffusion, (2) direct thermo-chemical

Mikrostruktur undersøgelser af PVD Al-Ni diffusion ved 610C indikerer at diffusion

Udover at der er AlNi3i korngrænserne af aluminium strukturen, observeres der Al3Ni og

Med henblik på at reducere produktionsomkostningerne i forbindelse med fremstilling af

II. C. K. Kjartansdóttir, M. Caspersen, S. Egelund and P. Møller, “Electrochemical

III. C. K. Kjartansdóttir, A. Hossein, T. Kasama and P. Møller, “Investigations of the

List of Participating Activities

III. C. K. Kjartansdóttir and P. Møller, “Development of hydrogen electrodes for

IV. L. Yde, C. K. Kjartansdóttir, F. Allebrod, M. B. Mogensen, P. Møller, L. R. Hilbert, P.

V. A. B. G. S. W, “Ny produktionsmetode muliggør “stinkende billig” brint,“

3.2 High temp alkaline water electrolysis (HTAWE) 27

4.2 Composition and phase analysis 31

7.2 Hot dip aluminising 58

7.3 Thermo‐chemical diffusion of aluminium and nickel sheets 65

7.4 Aluminium ionic liquid electroplating 67

7.5 Physical vapour deposition of aluminium onto electroplated sulfamate nickel 71

7.6 Conclusions on the manufacturing of a high surface area nickel coating 73

8.4 Conclusions for efficiency and durability testing 86

According to the World Meteorological Organization (WMO), the concentration of

1.2 Objectives of the present study and structure of the thesis

Expected results from the project were as follows:

A generation of alkaline electrolysers characterized by:

The thesis is divided in to 9 chapters and 4 appended papers.

P-2: Electrochemical investigation of surface area effects on PVD Al-Ni as

P-3: Investigations of the diffusion mechanism of PVD Al and Ni couples at 610°C.

2 Fundamentals of Water Electrolysis

Fig. 2: Outline of the operating principles of an alkaline water electrolysis cell.

285.9 kJ⁄mol 237.2 kJ⁄mol 48.7 kJ⁄mol (7)

In Fig. 3 the benefits of producing at elevated temperatures is evident. By increasing the

2.3 The resistance in the electrolysis cell

 Resistance overpotential (ηres)

The activation overpotential represents the activation energies of the electrochemical

The hydrogen reaction in an alkaline media is widely accepted to be a combination of the

Fig. 5: Elementary reaction steps of HER in alkaline media.

Fig. 6: The deependence of th

As mentionedd before the actual

Thhe mechanism m for the oxygen evolutiion reaction (OER) is mo

3 Water Electrolysis Technologies

Manufacturer Technology Production Power Energy Efficiencya Max, H2 Location

n.a: information not available

3.1 Alkaline Water Electrolysis

Technology Cathode (HER) Anode (OER) Charge carrier

3.1.1 Cell components

3.1.1.1 The diaphragm

The criteria for the diaphragms are:

 Permeable for hydroxide ions and water

3.1.1.2 The electrodes

Composition Preparation Working Electrolyte Current ηhydrogen Publication

Composition Preparation Working Electrolyte Current ηhydrogen Publication

3.1.1.3 The electrolyte

which for a strong base such as KOH can be written as:

3.1.2 Cell configuration

3.1.2.1 Bipolar vs. monopolar

3.1.2.2 Zero‐gap and non‐zero‐gap design

3.2 High temperature alkaline water electrolysis (HTAWE)

PE EM electrolyysis is constrructed in a siimilar way as

Fig. 12: Outline